Datasets:

---

LLMDH

/

OpenScience

like 0

Follow

LLM - Digital Humanities 16

ERROR: type should be string, got "https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Evolution of the Amundsen Sea Polynya, Antarctica, 2016-2021 \n1 \nGrant J. Macdonald, Stephen F. Ackley and Alberto M. Mestas-Nuñez \n2 \nNASA Center for Advanced Measurements in Extreme Environments (CAMEE), University of Texas at San Anto-\n3 \nnio, San Antonio, TX 78249, USA \n4 \nCorrespondence to: Grant J. Macdonald (grant.macdonald@utsa.edu) \n5 \n6 Abstract. Polynyas are key sites of ice production during the winter and are important sites of biological activity \n7 \nand carbon sequestration during the summer. The Amundsen Sea Polynya (ASP) is the fourth largest Antarctic po-\n8 \nlynya, has recorded the highest primary productivity and lies in an embayment of key oceanographic significance. 9 \nHowever, knowledge of its dynamics, and of sub-annual variations in its area and ice production, is limited. In this \n10 \nstudy we primarily utilize Sentinel-1 SAR imagery, sea ice concentration products and climate reanalysis data, along \n11 \nwith bathymetric data, to analyze the ASP over the period November 2016 - March 2021. Specifically, we analyze \n12 \n(i) qualitative changes in the ASP’s characteristics and dynamics, and quantitative changes in (ii) summer polynya \n13 \narea, (iii) winter polynya area and ice production. From our analysis of SAR imagery we find that ice produced by \n14 \nthe ASP becomes stuck in the vicinity of the polynya and sometimes flows back into the polynya, contributing to its \n15 \nclosure and limiting further ice production. The polynya forms westward off a persistent chain of grounded icebergs \n16 \nthat are located at the site of a bathymetric high. Grounded icebergs also influence the outflow of ice and facilitate \n17 \nthe formation of a ‘secondary polynya’ at times. Additionally, unlike some polynyas, ice produced by the polynya \n18 \nflows westward after formation, along the coast and into the neighboring sea sector. During the summer and early \n19 \nwinter, broader regional sea ice conditions can play an important role in the polynya. The polynya opens in all sum-\n20 \nmers, but record-low sea ice conditions in 2016/17 cause it to become part of the open ocean. During the winter, an \n21 \naverage of 78% of ice production occurs in April-May and September-October, but large polynya events often asso-\n22 \nciated with high winds can cause ice production throughout the winter. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. For example, the Ross Ice Shelf polynya is estimated to produce several cubic kil-\n58 \nometers of ice annually, and along with the McMurdo Sound Polynya, may produce 20-50% of total sea ice in the \n59 \nregion (Drucker et al., 2011). The polynya-produced ice then forms part of the icepack, contributing to its character \n60 \nand potentially further thickening due to deformation. For example, ice formed by the Terra Nova Bay polynya in \n61 \nthe Ross Sea had a mean thickness 3-4 times that of the central Ross Sea, with 80% of the study area’s ice contained \n62 \nin deformed ice and ridges (Rack et al., 2020). Consequently, understanding of polynya evolution through the winter \n63 \nis important for understanding ice production and sea ice characteristics in the Southern Ocean. 64 \n \nThe Amundsen Sea Polynya (ASP), West Antarctica and the embayment in which it lies are of particular \n65 \ninterest for several reasons. The polynya is situated in the embayment into which the Thwaites and Pine Island Glac-\n66 \niers terminate and undergo ocean-driven melting, making the oceanography of the embayment of special interest \n67 \n(IMBIE team, 2018; Rignot et al., 2019). The ASP is also known to be a key site of primary productivity in the sum-\n68 \nmer, supporting rates of net primary production up to 2.5gC m−2 day−1, the highest for any Antarctic polynya (Arrigo \n69 \nand Van Dijken, 2003; Arrigo et al., 2012). Additionally, the ASP has been highlighted as an important site for ice \n70 \nproduction. It has been identified as the fourth highest polynya in Antarctica in terms of area and ice production, \n71 \nonly behind the Ross Ice Shelf Cape Darnley and Mertz polynyas (Tamura et al 2008; 2016; Nihashi and\n72 Coastal polynyas, or ‘latent heat polynyas’ (and henceforth referred to simply as ‘polynyas’), are sites of \n38 \nopen water surrounded by sea ice and land, glacier ice or fast ice (Armstrong, 1972; Tamura et al., 2008; Park et al., \n39 \n2018). These polynyas are distributed around the coast of Antarctica and are typically at fixed geographic locations \n40 \neach year. They develop because the ice that forms at these sites is regularly driven away by winds or ocean cur-\n41 \nrents, creating an opening in the icepack (Bromwich and Kurtz, 1984; Bromwich et al., 1993; 1998; Morales \n42 \nMaqueda et al. 2004; Sansiviero et al., 2017). https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 43 Between the summer months of approximately November and March these open water sites tend to remain \n44 \npersistently ice-free. Among other factors, the combination of ice-free conditions, summer sunlight, and the availa-\n45 \nbility of dissolved iron (e.g. Arrigo et al., 2008a; 2012; St-Laurent et al., 2017), enables large phytoplankton blooms \n46 \nto develop in polynyas during this summer period. These phytoplankton blooms fix carbon from dissolved carbon \n47 \ndioxide, some of which then sinks below the surface layer (Sweeney et al., 2003). As a result, the evolution of po-\n48 \nlynyas during the summer is considered a key factor in the primary productivity of the Southern Ocean, and conse-\n49 \nquently, also their role in the sequestration of Carbon Dioxide (the ‘biological pump’) (Arrigo et al., 2008b). 50 Between the winter months of approximately April and October polynyas tend to intermittently open and \n51 \nare smaller in area than in the summer. When a polynya does open during the winter, excess ocean heat is lost and \n52 \nnew sea ice quickly produced. Winds (usually katabatic winds) or ocean currents then push the newly produced sea \n53 \nice away and open the polynya again, producing yet more sea ice in the open area. Repeated polynya ‘events’ pro-\n54 \nduce new sea ice throughout the winter period and hence polynyas have been termed ‘factories’ of sea ice produc-\n55 \ntion (Kimura and Wakatsuchi, 2004; Assmann et al., 2005). Overall, polynyas are estimated to contribute around \n56 \n10% of all Antarctic sea ice cover (Tamura et al., 2008; Nihashi and Oshima, 2015). Regionally, polynyas can play \n57 \nan even larger role in production. For example, the Ross Ice Shelf polynya is estimated to produce several cubic kil-\n58 \nometers of ice annually, and along with the McMurdo Sound Polynya, may produce 20-50% of total sea ice in the \n59 \nregion (Drucker et al., 2011). The polynya-produced ice then forms part of the icepack, contributing to its character \n60 \nand potentially further thickening due to deformation. For example, ice formed by the Terra Nova Bay polynya in \n61 \nthe Ross Sea had a mean thickness 3-4 times that of the central Ross Sea, with 80% of the study area’s ice contained \n62 \nin deformed ice and ridges (Rack et al., 2020). https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Repeated polynya ‘events’ pro-\n54 \nduce new sea ice throughout the winter period and hence polynyas have been termed ‘factories’ of sea ice produc-\n55 \ntion (Kimura and Wakatsuchi, 2004; Assmann et al., 2005). Overall, polynyas are estimated to contribute around \n56 \n10% of all Antarctic sea ice cover (Tamura et al., 2008; Nihashi and Oshima, 2015). Regionally, polynyas can play \n57 \nan even larger role in production. For example, the Ross Ice Shelf polynya is estimated to produce several cubic kil-\n58 \nometers of ice annually, and along with the McMurdo Sound Polynya, may produce 20-50% of total sea ice in the \n59 \nregion (Drucker et al., 2011). The polynya-produced ice then forms part of the icepack, contributing to its character \n60 \nand potentially further thickening due to deformation. For example, ice formed by the Terra Nova Bay polynya in \n61 \nthe Ross Sea had a mean thickness 3-4 times that of the central Ross Sea, with 80% of the study area’s ice contained \n62 \nin deformed ice and ridges (Rack et al., 2020). Consequently, understanding of polynya evolution through the winter \n63 \nis important for understanding ice production and sea ice characteristics in the Southern Ocean. 64 \n \nThe Amundsen Sea Polynya (ASP), West Antarctica and the embayment in which it lies are of particular \n65 \ninterest for several reasons. The polynya is situated in the embayment into which the Thwaites and Pine Island Glac-\n66 \niers terminate and undergo ocean-driven melting, making the oceanography of the embayment of special interest \n67 \n(IMBIE team, 2018; Rignot et al., 2019). The ASP is also known to be a key site of primary productivity in the sum-\n68 \nmer, supporting rates of net primary production up to 2.5gC m−2 day−1, the highest for any Antarctic polynya (Arrigo \n69 \nand Van Dijken, 2003; Arrigo et al., 2012). Additionally, the ASP has been highlighted as an important site for ice \n70 \nproduction. It has been identified as the fourth highest polynya in Antarctica in terms of area and ice production, \n71 \nonly behind the Ross Ice Shelf, Cape Darnley, and Mertz polynyas (Tamura et al., 2008; 2016; Nihashi and \n72 \nOhshima, 2015, Nihashi et al., 2017). 73 1. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Introduction \n37 \n \nCoastal polynyas, or ‘latent heat polynyas’ (and henceforth referred to simply as ‘polynyas’), are sites of \n38 \nopen water surrounded by sea ice and land, glacier ice or fast ice (Armstrong, 1972; Tamura et al., 2008; Park et al., \n39 \n2018). These polynyas are distributed around the coast of Antarctica and are typically at fixed geographic locations \n40 \neach year. They develop because the ice that forms at these sites is regularly driven away by winds or ocean cur-\n41 \nrents, creating an opening in the icepack (Bromwich and Kurtz, 1984; Bromwich et al., 1993; 1998; Morales \n42 \nMaqueda et al. 2004; Sansiviero et al., 2017). 43 \n \nBetween the summer months of approximately November and March these open water sites tend to remain \n44 \npersistently ice-free. Among other factors, the combination of ice-free conditions, summer sunlight, and the availa-\n45 \nbility of dissolved iron (e.g. Arrigo et al., 2008a; 2012; St-Laurent et al., 2017), enables large phytoplankton blooms \n46 \nto develop in polynyas during this summer period. These phytoplankton blooms fix carbon from dissolved carbon \n47 \ndioxide, some of which then sinks below the surface layer (Sweeney et al., 2003). As a result, the evolution of po-\n48 \nlynyas during the summer is considered a key factor in the primary productivity of the Southern Ocean, and conse-\n49 \nquently, also their role in the sequestration of Carbon Dioxide (the ‘biological pump’) (Arrigo et al., 2008b). 50 \n \nBetween the winter months of approximately April and October polynyas tend to intermittently open and \n51 \nare smaller in area than in the summer. When a polynya does open during the winter, excess ocean heat is lost and \n52 \nnew sea ice quickly produced. Winds (usually katabatic winds) or ocean currents then push the newly produced sea \n53 \nice away and open the polynya again, producing yet more sea ice in the open area. Repeated polynya ‘events’ pro-\n54 \nduce new sea ice throughout the winter period and hence polynyas have been termed ‘factories’ of sea ice produc-\n55 \ntion (Kimura and Wakatsuchi, 2004; Assmann et al., 2005). Overall, polynyas are estimated to contribute around \n56 \n10% of all Antarctic sea ice cover (Tamura et al., 2008; Nihashi and Oshima, 2015). Regionally, polynyas can play \n57 \nan even larger role in production. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. While passive microwave data or daily sea \n23 \nice concentration products remain key for analyzing variations in polynya area and ice production, we find that the \n24 \nability to directly observe and qualitatively analyze the polynya at a high temporal and spatial resolution with Senti-\n25 \nnel-1 imagery provides important insights about the behavior of the polynya that are not possible with those da-\n26 \ntasets. 27 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 1. Introduction \n37 \n \nCoastal polynyas, or ‘latent heat polynyas’ (and henceforth referred to simply as ‘polynyas’), are sites of \n38 \nopen water surrounded by sea ice and land, glacier ice or fast ice (Armstrong, 1972; Tamura et al., 2008; Park et al., \n39 \n2018). These polynyas are distributed around the coast of Antarctica and are typically at fixed geographic locations \n40 \neach year. They develop because the ice that forms at these sites is regularly driven away by winds or ocean cur-\n41 \nrents, creating an opening in the icepack (Bromwich and Kurtz, 1984; Bromwich et al., 1993; 1998; Morales \n42 \nMaqueda et al. 2004; Sansiviero et al., 2017). 43 \n \nBetween the summer months of approximately November and March these open water sites tend to remain \n44 \npersistently ice-free. Among other factors, the combination of ice-free conditions, summer sunlight, and the availa-\n45 \nbility of dissolved iron (e.g. Arrigo et al., 2008a; 2012; St-Laurent et al., 2017), enables large phytoplankton blooms \n46 \nto develop in polynyas during this summer period. These phytoplankton blooms fix carbon from dissolved carbon \n47 \ndioxide, some of which then sinks below the surface layer (Sweeney et al., 2003). As a result, the evolution of po-\n48 \nlynyas during the summer is considered a key factor in the primary productivity of the Southern Ocean, and conse-\n49 \nquently, also their role in the sequestration of Carbon Dioxide (the ‘biological pump’) (Arrigo et al., 2008b). 50 \n \nBetween the winter months of approximately April and October polynyas tend to intermittently open and \n51 \nare smaller in area than in the summer. When a polynya does open during the winter, excess ocean heat is lost and \n52 \nnew sea ice quickly produced. Winds (usually katabatic winds) or ocean currents then push the newly produced sea \n53 \nice away and open the polynya again, producing yet more sea ice in the open area. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Consequently, understanding of polynya evolution through the winter \n63 \nis important for understanding ice production and sea ice characteristics in the Southern Ocean. 64 The Amundsen Sea Polynya (ASP), West Antarctica and the embayment in which it lies are of particular \n65 \ninterest for several reasons. The polynya is situated in the embayment into which the Thwaites and Pine Island Glac-\n66 \niers terminate and undergo ocean-driven melting, making the oceanography of the embayment of special interest \n67 \n(IMBIE team, 2018; Rignot et al., 2019). The ASP is also known to be a key site of primary productivity in the sum-\n68 \nmer, supporting rates of net primary production up to 2.5gC m−2 day−1, the highest for any Antarctic polynya (Arrigo \n69 \nand Van Dijken, 2003; Arrigo et al., 2012). Additionally, the ASP has been highlighted as an important site for ice \n70 \nproduction. It has been identified as the fourth highest polynya in Antarctica in terms of area and ice production, \n71 \nonly behind the Ross Ice Shelf, Cape Darnley, and Mertz polynyas (Tamura et al., 2008; 2016; Nihashi and \n72 \nOhshima, 2015, Nihashi et al., 2017). 73 2 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. While there have been several recent studies of the ASP’s evolution during the summer months (e.g. Arrigo \n74 \net al., 2012; Stammerjohn et al., 2015, St-Laurent et al., 2019), knowledge of the ASP and its role in ice production \n75 \nduring the winter is limited. Additionally, few studies during the summer have analyzed changes at the sub-monthly \n76 \nscale, and none have observed the polynya directly during cloudy conditions. Aside from one study that analyzed the \n77 \nASP at the mean monthly scale (Tamura et al., 2016), studies that analyze ice production in the ASP during the win-\n78 \nter have been limited to estimates of total annual ice production and mean annual area as part of broader-scale cir-\n79 \ncum-Antarctic studies (Tamura et al., 2008; Nihashi and Ohshima, 2015; Nihashi et al., 2017). Other studies of the \n80 \nASP during the winter have focused on other aspects of the polynya, such as iron and carbon fluxes (St-Laurent et \n81 \nal., 2019). There are a lack of studies of the polynya that characterize changes in the polynya’s evolution and area \n82 \nthrough individual seasons. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. This is partly due to the difficulty of analyzing polynyas in detail during the polar night. 83 \nHowever, the launch of the Sentinel-1 constellation of Synthetic Aperture Radar (SAR) - in full operation by May \n84 \n2016 - enables us to directly observe the polynya during the polar night at a high spatial resolution. Additionally, \n85 \nduring the summer light, SAR allows us to make observations regardless of cloud cover. 86 The overall goal of the work presented here is to improve knowledge of the behavior and evolution of the \n87 \nASP, and thus to aid understanding of recent complex and poorly understood trends in Southern Ocean sea ice con-\n88 \nditions. This in turn, will aid predictions of future changes in Southern Ocean sea condition due to climate change, \n89 \nwith important consequences for a range of processes, such as Antarctic bottom water formation and global thermo-\n90 \nhaline circulation (Orsi et al., 1999; Krumpen et al., 2011; Tamura et al., 2012; Ohshima et al., 2013; Kitade et al., \n91 \n2014; Marzocchi and Jansen, 2019), Antarctic Ice Sheet stability (Banwell et al., 2017; Webber et al., 2017; Greene \n92 \net al., 2018; Massom et al., 2018; Arthur et al., 2021) and ecosystem productivity (Grossman and Dieckmann, 1994; \n93 \nIto et al., 2017). The three specific objectives of this paper are to, over the period November 2016 - March 2021, \n94 \nanalyze seasonal and inter-annual (i) qualitative changes in the ASP’s characteristics and dynamics, and quantitative \n95 \nchanges in (ii) summer polynya area and (iii) winter polynya area and ice production. The main datasets used are \n96 \nSentinel-1 SAR images, sea ice concentration products, and climate reanalysis data in the region of the ASP. Addi-\n97 \ntionally, we analyze bathymetric data, and changes in the broader regional sea ice. 98 \n99 2. Study Site Polynya opening in November was \n119 \nassociated with prevailing easterly or southeasterly winds, while closure in March was associated with persistent \n120 \nsoutheasterly winds at a time when winds promote ice growth in open areas. The polynya was also found to open 16 \n121 \n ±7 days earlier at the end of the period 1979/80-2013/14 than the beginning (Stammerjohn et al., 2015). 122 \n \nDuring the winter, the polynya’s area was estimated to have a daily mean of 7700 ± 3600 km2 for the pe-\n123 \nriod March-October, 2003-11, as estimated from Advanced Microwave Scanning Radiometer for EOS (AMSR-E) \n124 \ndata (Nihashi and Ohshima, 2015). Annual ice production has been estimated as 92 ± 16 km3 for the period 1992-\n125 \n2001 (Tamura et al., 2008) and 123 ± 24 km3 for the period 1992-2013 (Tamura et al., 2016) using Special Sensor \n126 \nMicrowave/Imager (SSM/I) data. Nihashi et al., 2017 estimated annual ice production as 90 ± 13 km3 (AMSR-E \n127 \ndata) for the period 2003-10, and 90 ± 17 km3 (AMSR2) for the period 2013-15. 128 The ASP opened every summer during the period 1979-2014 studied by Stammerjohn et al. (2015) and re-\n110 \ntained some open polynya area through the winter period. Arrigo et al. (2012) found no significant secular trend in \n111 \nmean summer open water area between 1997 and 2010, but Stammerjohn et al. (2015) did find the ASP’s area in \n112 \nDecember-February to increase overall over the period 1979-2014. They also noted that the site of the polynya \n113 \nopening shifted to its current typical site adjacent to the Thwaites Glacier Tongue in 1993, having previously been \n114 \nfurther to the west. 115 \n \nSynoptic-scale winds have been found to primarily determine the ASP’s area and the timing of opening and \n116 \nclosure. Over the period 1997-2010, ASP area was greatest in the summers of 2002-03 and 2009-10, the years with \n117 \nthe largest monthly anomalies in easterly and southerly surface winds in the region, and smallest in 2003-04 when \n118 \nthere were anomalously high northerly and westerly winds (Arrigo et al., 2012). Polynya opening in November was \n119 \nassociated with prevailing easterly or southeasterly winds, while closure in March was associated with persistent \n120 \nsoutheasterly winds at a time when winds promote ice growth in open areas. 2. Study Site The polynya was also found to open 16 \n121 \n ±7 days earlier at the end of the period 1979/80-2013/14 than the beginning (Stammerjohn et al., 2015). 122 \n \nDuring the winter, the polynya’s area was estimated to have a daily mean of 7700 ± 3600 km2 for the pe-\n123 \nriod March-October, 2003-11, as estimated from Advanced Microwave Scanning Radiometer for EOS (AMSR-E) \n124 \ndata (Nihashi and Ohshima, 2015). Annual ice production has been estimated as 92 ± 16 km3 for the period 1992-\n125 \n2001 (Tamura et al., 2008) and 123 ± 24 km3 for the period 1992-2013 (Tamura et al., 2016) using Special Sensor \n126 \nMicrowave/Imager (SSM/I) data. Nihashi et al., 2017 estimated annual ice production as 90 ± 13 km3 (AMSR-E \n127 \ndata) for the period 2003-10, and 90 ± 17 km3 (AMSR2) for the period 2013-15. 128 The ASP opened every summer during the period 1979-2014 studied by Stammerjohn et al. (2015) and re-\n110 \ntained some open polynya area through the winter period. Arrigo et al. (2012) found no significant secular trend in \n111 \nmean summer open water area between 1997 and 2010, but Stammerjohn et al. (2015) did find the ASP’s area in \n112 \nDecember-February to increase overall over the period 1979-2014. They also noted that the site of the polynya \n113 \nopening shifted to its current typical site adjacent to the Thwaites Glacier Tongue in 1993, having previously been \n114 \nfurther to the west. 115 Synoptic-scale winds have been found to primarily determine the ASP’s area and the timing of opening and \nclosure. Over the period 1997-2010, ASP area was greatest in the summers of 2002-03 and 2009-10, the years with \nthe largest monthly anomalies in easterly and southerly surface winds in the region, and smallest in 2003-04 when \nthere were anomalously high northerly and westerly winds (Arrigo et al., 2012). Polynya opening in November was \nassociated with prevailing easterly or southeasterly winds, while closure in March was associated with persistent \nsoutheasterly winds at a time when winds promote ice growth in open areas. The polynya was also found to open 16 \n ±7 days earlier at the end of the period 1979/80-2013/14 than the beginning (Stammerjohn et al., 2015). 2. Study Site The ASP is located at around ~72-73°S and 110-120°W in the Amundsen Sea embayment of the Southern \n101 \nOcean in West Antarctica (Fig. 1). It is situated in a sector that exhibited an anomalous 40-year decreasing trend in \n102 \nsea ice extent until a 2007 minimum, since which there has been an increasing trend (Parkinson, 2019). To the east, \n103 \nthe polynya is bound by the Thwaites Iceberg Tongue and a chain of icebergs. To the south, when at its maximum \n104 \nextent, the polynya abuts the Dotson Ice Shelf and part of the Getz Ice Shelf. Immediately east of the eastern bound-\n105 \nary of the polynya is an area of ocean that is adjacent to Thwaites Glacier and Pine Island Glacier. The neighboring \n106 \n‘Pine Island Polynya’ forms along the coastal stretch around this area and to the north. Westward coastal currents \n107 \nprevail in the area (St-Laurent et al., 2019), that, along with easterly winds, carry icebergs (Koo et al., in review) and \n108 \nsea ice into the adjacent sector or the Amundsen Sea and eventually to the Ross Sea (Assmann et al., 2005). 109 3 3 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. The ASP opened every summer during the period 1979-2014 studied by Stammerjohn et al. (2015) and re-\n110 \ntained some open polynya area through the winter period. Arrigo et al. (2012) found no significant secular trend in \n111 \nmean summer open water area between 1997 and 2010, but Stammerjohn et al. (2015) did find the ASP’s area in \n112 \nDecember-February to increase overall over the period 1979-2014. They also noted that the site of the polynya \n113 \nopening shifted to its current typical site adjacent to the Thwaites Glacier Tongue in 1993, having previously been \n114 \nfurther to the west. 115 \n \nSynoptic-scale winds have been found to primarily determine the ASP’s area and the timing of opening and \n116 \nclosure. Over the period 1997-2010, ASP area was greatest in the summers of 2002-03 and 2009-10, the years with \n117 \nthe largest monthly anomalies in easterly and southerly surface winds in the region, and smallest in 2003-04 when \n118 \nthere were anomalously high northerly and westerly winds (Arrigo et al., 2012). 3.1 Qualitative analysis of the ASP’s evolution \n137 In order to qualitatively characterize the seasonal and interannual evolution of the ASP we use Sentinel-1 \n138 \nSAR imagery. Sentinel-1 is a constellation of two satellites, A and B, that were launched by the European Space \n139 \nAgency (ESA) in 2014 and 2016, respectively. The satellite collects radar backscatter imagery in the C-band which \n140 \nallows observations of sea ice and the ocean during cloudy conditions and the polar night. 141 For our analysis we processed all Sentinel-1 extra-wide swath (EW) mode, Ground Range Detected (GRD) \n142 \nimages over the study site and its surroundings (Fig. 1b) for the period November 2016 to March 2021. This period \n143 \nwas chosen because it includes all the complete summer (November-March) and winter (April-October) periods dur-\n144 \ning which both satellites A and B of the Sentinel-1 SAR constellation have been active. The EW mode was primar-\n145 \nily designed for sea ice and polar zones and collects images over a wider area than other modes. EW images are \n146 \navailable in 20m x 40m spatial resolution and all images were resampled to 40 m grid spacing. Of four available \n147 \nband combinations (VV, HH, VV+VH, and HH+HV), we use the HH band because most of the images contain this \n148 \nband. Using these images, we created a time-lapse animation using Google Earth Engine. This time-lapse included \n149 \nat least partial coverage of the study area for 56 days in 2016, 359 days in 2017, 341 days in 2018, 317 days in 2019, \n150 \n329 days in 2020 and 85 days in 2021. In order to analyze particular images in detail, the images were also down-\n151 \nloaded from the Alaska Satellite Facility (asf.alaska.edu) and processed in ESA’s ‘SNAP’ toolbox. SNAP was used \n152 \nto crop the images, apply radiometric correction, speckle filtering and ellipsoid correction and convert the images to \n153 \ndecibel values. The images were then loaded into QGIS (QGIS.org, 2021) for analysis. 154 Qualitative analysis was carried out by visually analyzing the time-lapse videos and images of interest, not-\n155 \ning changes in the state of the polynya and ice in the region. Visual analysis is possible because of the distinct \n156 \nbackscatter signals and texture of open water and different types of sea ice. 2. Study Site Typically, open ocean water has a low \n157 \nbackscatter and appears dark, while thicker, older icepack has a relatively high backscatter and appears bright and \n158 \nmore granular (we refer to all ice not produced by the ASP as ‘icepack’) (Fig. 2a-b). Recently-formed polynya-pro-\n159 \nduced ice has an intermediate backscatter (Fig. 2a). Frazil ice, that may form when a polynya opens up and the open \n160 \nocean begins to freeze, forms in distinct bands of varying brightness (Fig. 2c-d). 161 \n \nGiven the role grounded icebergs play in bounding the ASP, we also downloaded the ‘BedMachine Antarc-\n162 \ntica V2’ sea floor topography dataset for our study area to examine alongside our qualitative analysis. This dataset \n163 \nwas downloaded from the NSIDC (https://nsidc.org/data/nsidc-0756) and has a grid spacing of 500 x 500 m \n164 \n(Morlighem et al., 2020). 165 \n \nWe also use our analysis of the imagery to assess the approximate day of summer polynya ‘opening’ and \n166 \n‘ l\ni\n’\nd\nh\nl\nb\nf\nh\nh\nh\nl\ni\ni\nil f\nf\ni\n167 Fig. 1. (a) The location of the Amundsen Sea and our study sites within the context of Antarctica and the Southern \n130 \nOcean. The background image is from Quantarctica (Matsuoka et al., 2021); (b) The location of the ASP within the \n131 \nAmundsen Sea embayment. The green boundary indicates the area defined as the ‘ASP study area’ for the purpose \n132 \nof calculating winter polynya area and ice production. The background image is a true-color MODIS image from 12 \n133 \nDecember 2020. 134 \n135 2. Study Site During the winter, the polynya’s area was estimated to have a daily mean of 7700 ± 3600 km2 for the pe-\n123 \nriod March-October, 2003-11, as estimated from Advanced Microwave Scanning Radiometer for EOS (AMSR-E) \n124 \ndata (Nihashi and Ohshima, 2015). Annual ice production has been estimated as 92 ± 16 km3 for the period 1992-\n125 \n2001 (Tamura et al., 2008) and 123 ± 24 km3 for the period 1992-2013 (Tamura et al., 2016) using Special Sensor \n126 \nMicrowave/Imager (SSM/I) data. Nihashi et al., 2017 estimated annual ice production as 90 ± 13 km3 (AMSR-E \n127 \ndata) for the period 2003-10, and 90 ± 17 km3 (AMSR2) for the period 2013-15. 128 \u0001\u0002\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0006\u0003\u0004\n\u0001\u0007\u0003\u0004\n\b\u0005\n\b\u0005\n\b\b\n\b\b\t\u0003\n\b\b\n\b\b\u0002\u0003\n\b\u0002\n\b\u0002\t\u0003\n\b\u0002\u0002\u0003\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\u0004\b\n\t\u0007\n\b\u000b\f\r\u000e\u0004\u000e\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\b\t\u0003\n\u0005\u000b\b\t\f\u0002\u0004\r\u0003\n\u0006\f\u0004\u0005\u000e\u000f\u0003\u0010\u0011\u0004\n\u000e\u0012\u0013\t\u0002\u0014\u0004\u0007\n\u000b\b\t\f\u0002\u0004\r\n\u0015\u000f\u0014\u0007\u000f\u0003\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u0018\u0019\u0019\u000f\u0014\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u000b\u0004\u0014\u001a\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u001b\u0019 \u0001\u0002\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0006\u0003\u0004\n\u0001\u0007\u0003\u0004\n\b\u0005\n\b\u0005\n\b\b\n\b\b\t\u0003\n\b\b\n\b\b\u0002\u0003\n\b\u0002\n\b\u0002\t\u0003\n\b\u0002\u0002\u0003\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\u0004\b\n\t\u0007\n\b\u000b\f\r\u000e\u0004\u000e\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\b\t\u0003\n\u0005\u000b\b\t\f\u0002\u0004\r\u0003\n\u0006\f\u0004\u0005\u000e\u000f\u0003\u0010\u0011\u0004\n\u000e\u0012\u0013\t\u0002\u0014\u0004\u0007\n\u000b\b\t\f\u0002\u0004\r\n\u0015\u000f\u0014\u0007\u000f\u0003\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u0018\u0019\u0019\u000f\u0014\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u000b\u0004\u0014\u001a\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u001b\u0019 129 \n\u0001\u0002\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0005\u0003\u0004\n\u0001\u0006\u0003\u0004\n\u0001\u0007\u0003\u0004\n\b\u0005\t\u0003\n\b\u0005\u0002\u0003\n\b\b\t\u0003\n\b\b\t\u0003\n\b\b\u0002\u0003\n\b\b\u0002\u0003\n\b\u0002\t\u0003\n\b\u0002\t\u0003\n\b\u0002\u0002\u0003\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\u0004\b\n\t\u0007\n\b\u000b\f\r\u000e\u0004\u000e\n\u0001\u0002\u0003\u0004\u0005\u0006\u0007\b\t\u0003\n\u0005\u000b\b\t\f\u0002\u0004\r\u0003\n\u0006\f\u0004\u0005\u000e\u000f\u0003\u0010\u0011\u0004\n\u000e\u0012\u0013\t\u0002\u0014\u0004\u0007\n\u000b\b\t\f\u0002\u0004\r\n\u0015\u000f\u0014\u0007\u000f\u0003\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u0018\u0019\u0019\u000f\u0014\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u000b\u0004\u0014\u001a\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017\n\u001b\u0019\u001c\n\u0001\u0002\u0003\u0003\u0004\n\u0005\u0006\u0007\n\u000f\u0007\u0010\n\u0011\u0012\u000e\u000b\u0013\u0003\u0006\u000b\n\u0005\u0006\u0007\n\u0014\u0015\u0016\u0006\u000b\u0016\u0004\u0002\u0017\u0004\n\u0018\n\f\u0019\u0004\u001a\u001b\n%HOOLQJVKDXVHQ\n6HD 129\n\u0001\u0002\u0003\u0003\u0004\n\u0005\u0006\u0007\n\u000f\u0007\u0010\n\u0011\u0012\u000e\u000b\u0013\u0003\u0006\u000b\n\u0005\u0006\u0007\n\u0014\u0015\u0016\u0006\u000b\u0016\u0004\u0002\u0017\u0004\n\u0018\n\f\u0019\u0004\u001a\u001b\n%HOOLQJVKDXVHQ\n6HD \u0018\u0019\u0019\u000f\u0014\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017 \u0004\u0014\u001a\n\u0006\f\u0004\u0005\u0016\u0012\u0004\b\u0017 \u0001\u0002\u0003\u0004 \u0001\u0005\u0003\u0004 129 4 4 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Fig. 1. (a) The location of the Amundsen Sea and our study sites within the context of Antarctica and the Southern \n130 \nOcean. The background image is from Quantarctica (Matsuoka et al., 2021); (b) The location of the ASP within the \n131 \nAmundsen Sea embayment. The green boundary indicates the area defined as the ‘ASP study area’ for the purpose \n132 \nof calculating winter polynya area and ice production. The background image is a true-color MODIS image from 12 \n133 \nDecember 2020. 134 \n \n135 \n3. 2. Study Site Data & Methods \n136 \n3.1 Qualitative analysis of the ASP’s evolution \n137 \n \nIn order to qualitatively characterize the seasonal and interannual evolution of the ASP we use Sentinel-1 \n138 \nSAR imagery. Sentinel-1 is a constellation of two satellites, A and B, that were launched by the European Space \n139 \nAgency (ESA) in 2014 and 2016, respectively. The satellite collects radar backscatter imagery in the C-band which \n140 \nallows observations of sea ice and the ocean during cloudy conditions and the polar night. 141 \n \nFor our analysis we processed all Sentinel-1 extra-wide swath (EW) mode, Ground Range Detected (GRD) \n142 \nimages over the study site and its surroundings (Fig. 1b) for the period November 2016 to March 2021. This period \n143 \nwas chosen because it includes all the complete summer (November-March) and winter (April-October) periods dur-\n144 \ning which both satellites A and B of the Sentinel-1 SAR constellation have been active. The EW mode was primar-\n145 \nily designed for sea ice and polar zones and collects images over a wider area than other modes. EW images are \n146 \navailable in 20m x 40m spatial resolution and all images were resampled to 40 m grid spacing. Of four available \n147 \nband combinations (VV, HH, VV+VH, and HH+HV), we use the HH band because most of the images contain this \n148 \nband. Using these images, we created a time-lapse animation using Google Earth Engine. This time-lapse included \n149 \nat least partial coverage of the study area for 56 days in 2016, 359 days in 2017, 341 days in 2018, 317 days in 2019, \n150 \n329 days in 2020 and 85 days in 2021. In order to analyze particular images in detail, the images were also down-\n151 \nloaded from the Alaska Satellite Facility (asf.alaska.edu) and processed in ESA’s ‘SNAP’ toolbox. SNAP was used \n152 \nto crop the images, apply radiometric correction, speckle filtering and ellipsoid correction and convert the images to \n153 \ndecibel values. The images were then loaded into QGIS (QGIS.org, 2021) for analysis. 154 \n \nQualitative analysis was carried out by visually analyzing the time-lapse videos and images of interest, not-\n155 \ning changes in the state of the polynya and ice in the region. Visual analysis is possible because of the distinct \n156 \nbackscatter signals and texture of open water and different types of sea ice. 3.1 Qualitative analysis of the ASP’s evolution \n137 Polynya area was then calcu-\n192 \nlated by defining any pixel in the study area with a SIC < 70% as being part of the open polynya. The 70% threshold \n193 \nhas been commonly used in other studies of polynyas (e.g. Parmiggiani, 2006; Morelli & Parmiggiani, 2013; \n194 \nPreußer et al., 2015). A limitation is that smaller areas of open water that are represented in a pixel dominated by \n195 \nice-covered area (i.e. > 70%) will not be included in our polynya area value, while ice-covered areas in pixels with \n196 \nSIC < 70% will be included. However, based on visual comparison with coincident Sentinel-1 imagery, we found \n197 \n70% to be an appropriate threshold for accurately estimating open polynya area\n198 After each day’s data was downloaded as a geotiff, it was cropped to the 70 660 km2 ASP study area de-\n191 \nfined in Fig. 1b using a shapefile drawn in QGIS with a Sentinel-1 image as reference. Polynya area was then calcu-\n192 \nlated by defining any pixel in the study area with a SIC < 70% as being part of the open polynya. The 70% threshold \n193 \nhas been commonly used in other studies of polynyas (e.g. Parmiggiani, 2006; Morelli & Parmiggiani, 2013; \n194 \nPreußer et al., 2015). A limitation is that smaller areas of open water that are represented in a pixel dominated by \n195 \nice-covered area (i.e. > 70%) will not be included in our polynya area value, while ice-covered areas in pixels with \n196 \nSIC < 70% will be included. However, based on visual comparison with coincident Sentinel-1 imagery, we found \n197 \n70% to be an appropriate threshold for accurately estimating open polynya area. 198 While Sentinel-1 imagery has been used to obtain polynya area during the polar night at a higher spatial \n199 \nresolution (40m, Dai et al., 2020), the Bremen SIC product has three key advantages over using Sentinel-1 SAR im-\n200 \nagery. First, the SIC product is available daily, in contrast to Sentinel-1 which has many, and sometimes prolonged \n201 \ndata gaps over the primary area of interest, particularly during June/July. Given that polynya area can change sub-\n202 \nstantially on a daily or hourly timescale, regular gaps of successive days significantly limits the ability to quantita-\n203 \ntively characterize variations throughout the year. 3.1 Qualitative analysis of the ASP’s evolution \n137 198 \n \nWhile Sentinel-1 imagery has been used to obtain polynya area during the polar night at a higher spatial \n199 \nresolution (40m, Dai et al., 2020), the Bremen SIC product has three key advantages over using Sentinel-1 SAR im-\n200 \nagery. First, the SIC product is available daily, in contrast to Sentinel-1 which has many, and sometimes prolonged \n201 \ndata gaps over the primary area of interest, particularly during June/July. Given that polynya area can change sub-\n202 \nstantially on a daily or hourly timescale, regular gaps of successive days significantly limits the ability to quantita-\n203 \ntively characterize variations throughout the year. Second, several Sentinel-1 images are required to capture the \n204 \nwhole ASP study area on a particular day meaning that even on many days where there are images that are useful\n205 169 \n3.2 Daily polynya area \n170 \n \nIn order to analyze seasonal and interannual changes in polynya area in summer and winter, daily sea ice \n171 \nconcentration (SIC) for the study region was downloaded from the University of Bremen’s sea ice data center \n172 \n(seaice.uni-bremen.de). The data was separated into five summer periods from November to March (2016/17, \n173 \n2017/18, 2018/19, 2019/20 2020/21) and four winter periods from April to October (2017, 2018, 2019, 2020). This \n174 \ntime period was focused on because it coincides with the period for which there is Sentinel-1 A and B data. We \n175 \nbegin our winter period in April rather than March because analysis of the Sentinel-1 imagery suggests ice produc-\n176 \ntion is not active across the open polynya at the beginning of March. The sea ice concentration product was pro-\n177 \ncessed by the University of Bremen using the ARTIST Sea Ice (ASIC) algorithm (Spreen et al, 2008) applied to \n178 \nAMSR-2 data. AMSR-2 was launched onboard the Japan Aerospace Exploration Agency’s (JAXA) Global Change \n179 \nObservation Mission - Water (GCOM-W) satellite in July 2012. 180 We used version 5.4 of the Antarctic-wide, daily sea ice concentration product with no land mask, pro-\n181 \ncessed to 3.125 km grid spacing. This is of a higher-resolution than data previously used to analyze polynya area in \n182 \nthe region. For example, Arrigo et al., (2012) used SSM/I data with 6.25 km grid spacing for their study of summer \n183 \npolynya area. Nihashi et al. 3.1 Qualitative analysis of the ASP’s evolution \n137 (2017) used AMSR-E data with 6.25 km grid spacing for their estimates of ice produc-\n184 \ntion. Tamura et al. (2008; 2016) also used SSM/I data, with 12.5 km grid spacing, for estimates of ice production. 185 \nStammerjohn et al. (2015) used Bootstrap SIC data with 25 km grid spacing for their analysis of summer polynya \n186 \narea. However, with our higher-resolution data (3.125 km) there remain limitations in using data with such a scale to \n187 \nmeasure something that can vary on a meterscale. It has been estimated that the ice concentration error in our \n188 \nAMSR-2 dataset is 25% at 0% SIC, decreasing to <10% error for SIC over 65% and 5.7% error at 100% SIC \n189 \n(Spreen et al., 2008). Data was available for all days in our study period apart from one day in 2019 (1 September). 190 We used version 5.4 of the Antarctic-wide, daily sea ice concentration product with no land mask, pro-\n181 \ncessed to 3.125 km grid spacing. This is of a higher-resolution than data previously used to analyze polynya area in \n182 \nthe region. For example, Arrigo et al., (2012) used SSM/I data with 6.25 km grid spacing for their study of summer \n183 \npolynya area. Nihashi et al. (2017) used AMSR-E data with 6.25 km grid spacing for their estimates of ice produc-\n184 \ntion. Tamura et al. (2008; 2016) also used SSM/I data, with 12.5 km grid spacing, for estimates of ice production. 185 \nStammerjohn et al. (2015) used Bootstrap SIC data with 25 km grid spacing for their analysis of summer polynya \n186 \narea. However, with our higher-resolution data (3.125 km) there remain limitations in using data with such a scale to \n187 \nmeasure something that can vary on a meterscale. It has been estimated that the ice concentration error in our \n188 \nAMSR-2 dataset is 25% at 0% SIC, decreasing to <10% error for SIC over 65% and 5.7% error at 100% SIC \n189 \n(Spreen et al., 2008). Data was available for all days in our study period apart from one day in 2019 (1 September). 190 \n \nAfter each day’s data was downloaded as a geotiff, it was cropped to the 70 660 km2 ASP study area de-\n191 \nfined in Fig. 1b using a shapefile drawn in QGIS with a Sentinel-1 image as reference. 3.1 Qualitative analysis of the ASP’s evolution \n137 Typically, open ocean water has a low \n157 \nbackscatter and appears dark, while thicker, older icepack has a relatively high backscatter and appears bright and \n158 \nmore granular (we refer to all ice not produced by the ASP as ‘icepack’) (Fig. 2a-b). Recently-formed polynya-pro-\n159 \nduced ice has an intermediate backscatter (Fig. 2a). Frazil ice, that may form when a polynya opens up and the open \n160 \nocean begins to freeze, forms in distinct bands of varying brightness (Fig. 2c-d). 161 Given the role grounded icebergs play in bounding the ASP, we also downloaded the ‘BedMachine Antarc-\n162 \ntica V2’ sea floor topography dataset for our study area to examine alongside our qualitative analysis. This dataset \n163 \nwas downloaded from the NSIDC (https://nsidc.org/data/nsidc-0756) and has a grid spacing of 500 x 500 m \n164 \n(Morlighem et al., 2020). 165 We also use our analysis of the imagery to assess the approximate day of summer polynya ‘opening’ and \n166 \n‘closing’. We deem the polynya to be open for the summer when the open polynya area is primarily free of active \n167 \nice production, and the day of summer closing to be when the whole open polynya is subject to ice production. 168 5 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 169 \n3.2 Daily polynya area \n170 \n \nIn order to analyze seasonal and interannual changes in polynya area in summer and winter, daily sea ice \n171 \nconcentration (SIC) for the study region was downloaded from the University of Bremen’s sea ice data center \n172 \n(seaice.uni-bremen.de). The data was separated into five summer periods from November to March (2016/17, \n173 \n2017/18, 2018/19, 2019/20 2020/21) and four winter periods from April to October (2017, 2018, 2019, 2020). This \n174 \ntime period was focused on because it coincides with the period for which there is Sentinel-1 A and B data. We \n175 \nbegin our winter period in April rather than March because analysis of the Sentinel-1 imagery suggests ice produc-\n176 \ntion is not active across the open polynya at the beginning of March. The sea ice concentration product was pro-\n177 \ncessed by the University of Bremen using the ARTIST Sea Ice (ASIC) algorithm (Spreen et al, 2008) applied to \n178 \nAMSR-2 data. AMSR-2 was launched onboard the Japan Aerospace Exploration Agency’s (JAXA) Global Change \n179 \nObservation Mission - Water (GCOM-W) satellite in July 2012. 3.1 Qualitative analysis of the ASP’s evolution \n137 180 \n \nWe used version 5.4 of the Antarctic-wide, daily sea ice concentration product with no land mask, pro-\n181 \ncessed to 3.125 km grid spacing. This is of a higher-resolution than data previously used to analyze polynya area in \n182 \nthe region. For example, Arrigo et al., (2012) used SSM/I data with 6.25 km grid spacing for their study of summer \n183 \npolynya area. Nihashi et al. (2017) used AMSR-E data with 6.25 km grid spacing for their estimates of ice produc-\n184 \ntion. Tamura et al. (2008; 2016) also used SSM/I data, with 12.5 km grid spacing, for estimates of ice production. 185 \nStammerjohn et al. (2015) used Bootstrap SIC data with 25 km grid spacing for their analysis of summer polynya \n186 \narea. However, with our higher-resolution data (3.125 km) there remain limitations in using data with such a scale to \n187 \nmeasure something that can vary on a meterscale. It has been estimated that the ice concentration error in our \n188 \nAMSR-2 dataset is 25% at 0% SIC, decreasing to <10% error for SIC over 65% and 5.7% error at 100% SIC \n189 \n(Spreen et al., 2008). Data was available for all days in our study period apart from one day in 2019 (1 September). 190 \n \nAfter each day’s data was downloaded as a geotiff, it was cropped to the 70 660 km2 ASP study area de-\n191 \nfined in Fig. 1b using a shapefile drawn in QGIS with a Sentinel-1 image as reference. Polynya area was then calcu-\n192 \nlated by defining any pixel in the study area with a SIC < 70% as being part of the open polynya. The 70% threshold \n193 \nhas been commonly used in other studies of polynyas (e.g. Parmiggiani, 2006; Morelli & Parmiggiani, 2013; \n194 \nPreußer et al., 2015). A limitation is that smaller areas of open water that are represented in a pixel dominated by \n195 \nice-covered area (i.e. > 70%) will not be included in our polynya area value, while ice-covered areas in pixels with \n196 \nSIC < 70% will be included. However, based on visual comparison with coincident Sentinel-1 imagery, we found \n197 \n70% to be an appropriate threshold for accurately estimating open polynya area. 3.1 Qualitative analysis of the ASP’s evolution \n137 Second, several Sentinel-1 images are required to capture the \n204 \nwhole ASP study area on a particular day, meaning that even on many days where there are images that are useful \n205 6 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. for qualitative analysis, the whole polynya cannot be measured. For example, in 2020 there is full coverage for only \n206 \n22 days with none between 26 April and 12 August. Third, even if sufficient images were available, current meth-\n207 \nods for calculating polynya area in Sentinel-1 imagery (e.g. Dai et al., 2020) requires manual delimitation which is \n208 \nlabor intensive and would be highly time consuming to do for multiple years at a daily temporal resolution. 209 \n \n210 \n3.3 Daily winter ice production \n211 \n \nIn order to calculate daily ice production in the ASP during the winter periods we applied a heat flux and \n212 \nice production model following Cheng et al. (2017). As input we used climate re-analysis data from the European \n213 \nCentre for Medium-Range Weather Forecasts Reanalysis v5 (ERA5) model and the same sea ice concentration data \n214 \nfrom the University of Bremen described in section 3.2. 215 \n \nHourly ERA5 data, with a spatial resolution of 31 km, was downloaded from Copernicus (cds.climate.co-\n216 \npernicus.eu; Herbach et al., 2018) for the following meteorological variables: air temperature at a height of 2 m, \n217 \nwind speed at a height of 10 m, surface air pressure, dewpoint temperature at a height of 2 m, downward solar radia-\n218 \ntion and downward thermal radiation. Air temperature, wind speed, surface air pressure and dewpoint temperature \n219 \nwere then processed to daily mean values, while solar and thermal radiation were processed to daily cumulative val-\n220 \nues. These calculations were done for the same ASP study site as for polynya area (Fig. 1b). 221 \n \n222 \n3.3.1 Heat Flux Calculation \n223 \n \nFollowing Cheng et al. 3.3 Daily winter ice production \n211 In order to calculate daily ice production in the ASP during the winter periods we applied a heat flux and \n212 \nice production model following Cheng et al. (2017). As input we used climate re-analysis data from the European \n213 \nCentre for Medium-Range Weather Forecasts Reanalysis v5 (ERA5) model and the same sea ice concentration data \n214 \nfrom the University of Bremen described in section 3.2. 215 3.1 Qualitative analysis of the ASP’s evolution \n137 (2017) the daily net heat flux, Q (in Wm−2), of a pixel was estimated by: \n224 \n \n225 \nQ = (1 − α)Si + Li − Lo + Fs + Fe (1) \n226 \n \n227 \nwhere Si (in Wm−2) is the cumulative downward solar radiation; Li (in Wm−2) is the cumulative downward thermal \n228 \nradiation; Lo (in Wm−2) is the upward thermal radiation; Fs (in Wm−2) and Fe (in Wm−2) are the sensible heat flux \n229 \nand latent heat flux, respectively; and α is the albedo of open water. α was taken to be 0.06 following Cheng et al. 230 \n(2017; 2019), Si and Li were taken from the processed daily ERA5 values and Lo, Fs and Fe were calculated as de-\n231 \nscribed below. 232 \n \nThe upward thermal radiation was calculated by the Stefan-Boltzmann law: \n233 \n \n234 \nLo = εσT0\n4\n (2) \n235 \n \n236 \nwhere ε is the longwave emissivity of open water (0.99), and σ is the Stefan–Boltzmann constant (5.67 × 10−8 \n237 \nW−2K−4). T0 (in K), the freezing point of seawater, was assumed to be the temperature of the water surface (TS, in K)\n238 \nwhich was calculated following Motoi et al. (1987) and Cheng et al. (2017; 2019) as: \n239 \n \n240 \nT0 = TS = 273.15 − 0.0137 − 0.05199sw − 0.00007225sw (3) \n241 \n \n242 for qualitative analysis, the whole polynya cannot be measured. For example, in 2020 there is full coverage for only \n206 \n22 days with none between 26 April and 12 August. Third, even if sufficient images were available, current meth-\n207 \nods for calculating polynya area in Sentinel-1 imagery (e.g. Dai et al., 2020) requires manual delimitation which is \n208 \nlabor intensive and would be highly time consuming to do for multiple years at a daily temporal resolution. 209 \n210 3.3.1 Heat Flux Calculation \n3 pixels considered as part of open polynya) daily ice production volume, V, was esti-\n275 \nmated in km3 following Cheng et al. (2017) by the following equation (9). Although ice production will also take \n276 \nplace in areas where there is ice cover, here we are only concerned with ice production taking place in the open po-\n277 \nlynya. 278 \n \n279 where sw (in ‰) is the salinity of sea water. The salinity of the Amundsen Sea was estimated as 34‰ based on Bett \n243 \net al. (2020). 244 \n \nThe sensible heat flux, (Fs), and latent heat flux, (Fe) was calculated by: \n245 \n \n246 \nFs = ρacpCsU(Ta − T0) (4) \n247 \n \n248 \nand \n249 \n \n250 \nFe = 0.622ρaLvCeU(rea − es)/P0 (5) \n251 \n \n252 \nwhere ρa is the density of air at standard atmospheric pressure and 0°C, taken as, 1.3 kg·m−3, cp is the specific heat \n253 \nof air at constant pressure, taken as 1004 J kg−1 K−1, U (in ms−1) is the wind speed at 10 m, taken from the processed \n254 \nERA5 data and Ta (in K) is the air temperature at 2 m, taken from the processed ERA5 data. Cs and Ce are bulk \n255 \ntransfer coefficients for sensible heat and latent heat, respectively and both taken as 0.00144. P0 (in pa) is the surface \n256 \nair pressure and taken from the processed ERA5 data. Lv (in J kg−1) is the latent heat of water vaporization, r is the \n257 \nrelative humidity. ea (in pa) is the saturation water vapor pressure at the air temperature, rea is the actual water vapor \n258 \npressure of the air and es (in pa) is the saturated water vapor pressure at the surface temperature and are all calcu-\n259 \nlated below: \n260 \n261 where sw (in ‰) is the salinity of sea water. The salinity of the Amundsen Sea was estimated as 34‰ based on Bett \n243 \net al. (2020). 244 where ρa is the density of air at standard atmospheric pressure and 0°C, taken as, 1.3 kg·m−3, cp is the specific heat \n253 \nof air at constant pressure, taken as 1004 J kg−1 K−1, U (in ms−1) is the wind speed at 10 m, taken from the processed \n254 \nERA5 data and Ta (in K) is the air temperature at 2 m, taken from the processed ERA5 data. 3.3.1 Heat Flux Calculation \n3 Following Cheng et al. (2017) the daily net heat flux, Q (in Wm−2), of a pixel was estimated by: Q = (1 − α)Si + Li − Lo + Fs + Fe (1) \n226 The upward thermal radiation was calculated by the Stefan-Boltzmann law: T0 = TS = 273.15 − 0.0137 − 0.05199sw − 0.00007225sw (3) \n241 7 7 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. where sw (in ‰) is the salinity of sea water. The salinity of the Amundsen Sea was estimated as 34‰ based on Bett \n243 \net al. (2020). 244 \n \nThe sensible heat flux, (Fs), and latent heat flux, (Fe) was calculated by: \n245 \n \n246 \nFs = ρacpCsU(Ta − T0) (4) \n247 \n \n248 \nand \n249 \n \n250 \nFe = 0.622ρaLvCeU(rea − es)/P0 (5) \n251 \n \n252 \nwhere ρa is the density of air at standard atmospheric pressure and 0°C, taken as, 1.3 kg·m−3, cp is the specific heat \n253 \nof air at constant pressure, taken as 1004 J kg−1 K−1, U (in ms−1) is the wind speed at 10 m, taken from the processed \n254 \nERA5 data and Ta (in K) is the air temperature at 2 m, taken from the processed ERA5 data. Cs and Ce are bulk \n255 \ntransfer coefficients for sensible heat and latent heat, respectively and both taken as 0.00144. P0 (in pa) is the surface \n256 \nair pressure and taken from the processed ERA5 data. Lv (in J kg−1) is the latent heat of water vaporization, r is the \n257 \nrelative humidity. ea (in pa) is the saturation water vapor pressure at the air temperature, rea is the actual water vapor \n258 \npressure of the air and es (in pa) is the saturated water vapor pressure at the surface temperature and are all calcu-\n259 \nlated below: \n260 \n \n261 \nLv = [2.501 − 0.00237(Ts − 273.15)] × 106 (6), \n262 \n \n263 \nes = 6.11.21 x 109.8094(T0-273.15)/(T0+0.71) (7), \n264 \n \n265 \nand \n266 \n \n267 \nrea =611.21×109.8094(Td−273.15)/(Td+0.71) (8) \n268 \n \n269 \nwhere Td is the dewpoint temperature taken from the processed ERA5 data. 270 \n \n271 \n3.3.2 Ice production calculation \n272 \n \nThe calculated daily heat flux was then cropped, re-aligned, resampled to a 3.125 km2 grid and reprojected \n273 \nto Antarctic Polar Stereographic using GDAL and QGIS to match the corresponding sea ice concentration data. 274 \nNext, where SIC was < 0.7 (i.e. 3.3.1 Heat Flux Calculation \n3 The daily data was plotted spatially for all available days 1 November 2016 - 31 \n299 \nMarch 2021, as shown in Video S2. Monthly mean SIC was also calculated for the whole period and plotted spa-\n300 \ntially. Additionally, the total SIC for each day was calculated by calculating the sum of all percentage SIC values in \n301 \nthe study region. These total SIC values should only be considered useful for analyzing relative changes in SIC in \n302 \nour study period. 303 \n \n304 \n3.5 Wind speed and direction \n305 \n \nIn order to analyze how polynya behavior relates to changes in wind conditions, mean wind speed and di-\n306 \nrection, and daily wind speed was calculated from ERA5 wind data. To obtain mean wind speed and direction, \n307 \nERA5’s monthly wind speed and direction product was downloaded and cropped to the ASP study area (Fig. 1b) for \n308 \nthe period 1 November 2016 to 31 December 2020 and the mean calculated for the whole period, and plotted spa-\n309 \ntially. 310 \n \nDaily wind speed and direction at the site of the polynya, ERA5’s hourly ‘u’ and ‘v’ wind products were \n311 \nprocessed for a region adjacent to the Dotson Ice Shelf and iceberg chain where the polynya typically forms, identi-\n312 \nfied in Fig. S1. Hourly wind speed in ms-1, V, was calculated as \n313 \n \n314 \nV = (u2 + v2)1/2 (11) \n315 3.3.1 Heat Flux Calculation \n3 Cs and Ce are bulk \n255 \ntransfer coefficients for sensible heat and latent heat, respectively and both taken as 0.00144. P0 (in pa) is the surface \n256 \nair pressure and taken from the processed ERA5 data. Lv (in J kg−1) is the latent heat of water vaporization, r is the \n257 \nrelative humidity. ea (in pa) is the saturation water vapor pressure at the air temperature, rea is the actual water vapor \n258 \npressure of the air and es (in pa) is the saturated water vapor pressure at the surface temperature and are all calcu-\n259 \nlated below: \n260 where Td is the dewpoint temperature taken from the processed ERA5 data. 270 8 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. V = 3.1252(1−SIC)Q/ρiLf (9) \n280 \n \n281 \nwhere, as above, SIC is sea ice concentration (as a fraction) and Q is daily net heat flux in W m−2, ρi is sea ice den-\n282 \nsity and taken as 920 kg m−3 and Lf is the latent heat of sea ice fusion in J kg−1. Lf is calculated following Moham-\n283 \nmed and Nirmal (2015) and Cheng et al. (2017) by: \n284 \n \n285 \nLf =333400−2113(T0 −273.15)−114si+18040si/(T0 −273.15)+3.35si(T0 −273.15)−3.76(T0−273.15)2 (10) \n286 \n \n287 \nwhere si is the salinity of sea ice, taken as 6‰ following Cheng et al. (2017). 288 \nCaution should be used when interpreting the absolute numbers produced by the ice production model, par-\n289 \nticularly because the input data is modeled climate data not necessarily always representative of reality, and the \n290 \nmodel itself is a simulation sensitive to uncertain parameter settings (Cheng et al., 2017; 2019). Nevertheless, we \n291 \nopted for this method due to the difficulty of directly measuring and tracking thin ice thickness in the polynya (e.g. 292 \nTian et al., 2020) to estimate ice production, and the potential to compare our daily ice production results to results \n293 \nobtained by the same model for the Ross Ice Shelf Polynya (Cheng et al., 2017; 2019). 294 \n \n295 \n3.4 Broader spatial changes in SIC \n296 \n \nIn order to assess changes in the ASP in the context of changes in SIC at a broader spatial scale, SIC was \n297 \nanalyzed for the larger area defined in Fig. 1a. The same SIC dataset described in section 2.2 to obtain polynya area \n298 \nwas cropped to the broader region. 3.4 Broader spatial changes in SIC \n296 In order to assess changes in the ASP in the context of changes in SIC at a broader spatial scale, SIC was \n297 \nanalyzed for the larger area defined in Fig. 1a. The same SIC dataset described in section 2.2 to obtain polynya area \n298 \nwas cropped to the broader region. The daily data was plotted spatially for all available days 1 November 2016 - 31 \n299 \nMarch 2021, as shown in Video S2. Monthly mean SIC was also calculated for the whole period and plotted spa-\n300 \ntially. Additionally, the total SIC for each day was calculated by calculating the sum of all percentage SIC values in \n301 \nthe study region. These total SIC values should only be considered useful for analyzing relative changes in SIC in \n302 \nour study period. 303 \n304 4. Results In this section, we first describe our qualitative analysis of the Sentinel-1 SAR imagery of the ASP between \n327 \nNovember 2016 and March 2021 (Section 4.1). Second, we analyze quantitative changes in summer (November - \n328 \nMarch) polynya area for the summers of 2016/17 to 2020/21 (Section 4.2). Third, we analyze quantitative changes \n329 \nin winter (April - October) polynya area and winter ice production for the winters of 2017 to 2020 (Section 4.3). 330 \nFourth, we analyze spatial and temporal variations in wind speed and how they relate to polynya area (Section 4.4), \n331 \nand finally, we analyze broader regional patterns in sea ice concentration for the period November 2016 - March \n332 \n2021 (Section 4.5). 333 3.5 Wind speed and direction \n305 Loss of most of the fast ice and icepack around the Iceberg Chain through the remainder of the summer \n341 \nmeans that only the presence of the Iceberg Chain delineates the boundary between the two polynyas for part of the \n342 \nsummer. Additionally, by early January the icepack that typically bounds the polynya to the north is no longer pre-\n343 \nsent, so the western area of the polynya has no northern limit and is congruent with the open ocean. 344 \n \nFrom around 13 February 2017 a narrow band of icepack from the north-west begins to drift around the \n345 \nIceberg Chain into the polynya, until by 3 March the polynya is once again isolated and bound to the west by the \n346 \nnarrow band of icepack. By late March new ice is forming in the polynya, until by approximately 4 April 2017 any \n347 \nopen polynya area appears to be the site of ice production, and we consider the summer phase of the polynya to be \n348 \n‘closed’ (Table 1). To the east and north-east ice production is also taking place in the Pine Island Polynya, and ice-\n349 \npack that has drifted from the Bellingshausen Sea fills the western side of the Pine Island Polynya, adjacent to the \n350 \nAbbott Ice Shelf. 351 \n \nFrom April 2017 onwards, numerous polynya events take place. During these events ice in the polynya typ-\n352 \nically blows to the west, creating an opening in the polynya adjacent to all or part of the Iceberg Chain. New ice can \n353 3.5 Wind speed and direction \n305 In order to analyze how polynya behavior relates to changes in wind conditions, mean wind speed and di-\ntion, and daily wind speed was calculated from ERA5 wind data. To obtain mean wind speed and direction, \nA5’s monthly wind speed and direction product was downloaded and cropped to the ASP study area (Fig. 1b) for \nperiod 1 November 2016 to 31 December 2020 and the mean calculated for the whole period, and plotted spa-\nly. Daily wind speed and direction at the site of the polynya, ERA5’s hourly ‘u’ and ‘v’ wind products were \n311 \nprocessed for a region adjacent to the Dotson Ice Shelf and iceberg chain where the polynya typically forms, identi-\n312 \nfied in Fig. S1. Hourly wind speed in ms-1, V, was calculated as \n313 9 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. where u and v are ERA5’s u and v 10 m wind products, respectively. Daily wind speed and direction was also plot-\n317 \nted spatially for the study area, included as supplementary video Video S3. For this, wind direction was plotted us-\n318 \ning the ‘matplotlib’ function ‘quiver’. 319 \n \nUnfortunately there is a lack of local observations of wind data, and the closest station in the United States \n320 \nAntarctic Program’s database, on Bear Peninsula, lacks wind data for most of our study period. As a result, some \n321 \ncaution should be employed when considering these results. However, Bracegirdle (2013) and Stammerjohn et al. 322 \n(2015) note that data from ERA5’s predecessor ERA-I in the neighboring Bellingshausen Sea performed better than \n323 \nother reanalysis products. 324 \n \n325 \n4. Results \n326 \n \nIn this section, we first describe our qualitative analysis of the Sentinel-1 SAR imagery of the ASP between \n327 \nNovember 2016 and March 2021 (Section 4.1). Second, we analyze quantitative changes in summer (November - \n328 \nMarch) polynya area for the summers of 2016/17 to 2020/21 (Section 4.2). Third, we analyze quantitative changes \n329 \nin winter (April - October) polynya area and winter ice production for the winters of 2017 to 2020 (Section 4.3). 3.5 Wind speed and direction \n305 330 \nFourth, we analyze spatial and temporal variations in wind speed and how they relate to polynya area (Section 4.4), \n331 \nand finally, we analyze broader regional patterns in sea ice concentration for the period November 2016 - March \n332 \n2021 (Section 4.5). 333 \n \n334 \n4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 \n \nBy the time of the first clear image of the ASP on 14 November 2016 the polynya is already open for the \n337 \nsummer, adjacent to the Iceberg Chain and the Dotson Ice Shelf, as well as part of the Getz Ice Shelf (Video S1). 338 \nThrough November and December, the polynya continues to extend westward. It also extends northward in the area \n339 \nwest of the Iceberg Chain. In late December and early January the Pine Island Polynya to the east has also extended \n340 \nwestward. Loss of most of the fast ice and icepack around the Iceberg Chain through the remainder of the summer \n341 \nmeans that only the presence of the Iceberg Chain delineates the boundary between the two polynyas for part of the \n342 \nsummer. Additionally, by early January the icepack that typically bounds the polynya to the north is no longer pre-\n343 \nsent, so the western area of the polynya has no northern limit and is congruent with the open ocean. 344 \n \nFrom around 13 February 2017 a narrow band of icepack from the north-west begins to drift around the \n345 \nIceberg Chain into the polynya, until by 3 March the polynya is once again isolated and bound to the west by the \n346 \nnarrow band of icepack. By late March new ice is forming in the polynya, until by approximately 4 April 2017 any \n347 \nopen polynya area appears to be the site of ice production, and we consider the summer phase of the polynya to be \n348 \n‘closed’ (Table 1). To the east and north-east ice production is also taking place in the Pine Island Polynya, and ice-\n349 \npack that has drifted from the Bellingshausen Sea fills the western side of the Pine Island Polynya, adjacent to the \n350 \nAbbott Ice Shelf. 351 \n \nFrom April 2017 onwards, numerous polynya events take place. 3.5 Wind speed and direction \n305 During these events ice in the polynya typ-\n352 \nically blows to the west, creating an opening in the polynya adjacent to all or part of the Iceberg Chain. New ice can \n353 where u and v are ERA5’s u and v 10 m wind products, respectively. Daily wind speed and direction was also plot-\n317 \nted spatially for the study area, included as supplementary video Video S3. For this, wind direction was plotted us-\n318 \ning the ‘matplotlib’ function ‘quiver’. 319 \n \nUnfortunately there is a lack of local observations of wind data, and the closest station in the United States \n320 \nAntarctic Program’s database, on Bear Peninsula, lacks wind data for most of our study period. As a result, some \n321 \ncaution should be employed when considering these results. However, Bracegirdle (2013) and Stammerjohn et al. 322 \n(2015) note that data from ERA5’s predecessor ERA-I in the neighboring Bellingshausen Sea performed better than \n323 \nother reanalysis products. 324 \n \n325 \n4. Results \n326 \n \nIn this section, we first describe our qualitative analysis of the Sentinel-1 SAR imagery of the ASP between \n327 \nNovember 2016 and March 2021 (Section 4.1). Second, we analyze quantitative changes in summer (November - \n328 \nMarch) polynya area for the summers of 2016/17 to 2020/21 (Section 4.2). Third, we analyze quantitative changes \n329 \nin winter (April - October) polynya area and winter ice production for the winters of 2017 to 2020 (Section 4.3). 330 \nFourth, we analyze spatial and temporal variations in wind speed and how they relate to polynya area (Section 4.4), \n331 \nand finally, we analyze broader regional patterns in sea ice concentration for the period November 2016 - March \n332 \n2021 (Section 4.5). 333 \n \n334 \n4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 \n \nBy the time of the first clear image of the ASP on 14 November 2016 the polynya is already open for the \n337 \nsummer, adjacent to the Iceberg Chain and the Dotson Ice Shelf, as well as part of the Getz Ice Shelf (Video S1). 338 \nThrough November and December, the polynya continues to extend westward. It also extends northward in the area \n339 \nwest of the Iceberg Chain. In late December and early January the Pine Island Polynya to the east has also extended \n340 \nwestward. 4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 By the time of the first clear image of the ASP on 14 November 2016 the polynya is already open for the \n7 \nsummer, adjacent to the Iceberg Chain and the Dotson Ice Shelf, as well as part of the Getz Ice Shelf (Video S1). 8 \nThrough November and December, the polynya continues to extend westward. It also extends northward in the area \n9 \nwest of the Iceberg Chain. In late December and early January the Pine Island Polynya to the east has also extended \n0 \nwestward. Loss of most of the fast ice and icepack around the Iceberg Chain through the remainder of the summer \n \nmeans that only the presence of the Iceberg Chain delineates the boundary between the two polynyas for part of the \n2 \nsummer. Additionally, by early January the icepack that typically bounds the polynya to the north is no longer pre-\n3 \nsent, so the western area of the polynya has no northern limit and is congruent with the open ocean. 4 From around 13 February 2017 a narrow band of icepack from the north-west begins to drift around the \n345 \nIceberg Chain into the polynya, until by 3 March the polynya is once again isolated and bound to the west by the \n346 \nnarrow band of icepack. By late March new ice is forming in the polynya, until by approximately 4 April 2017 any \n347 \nopen polynya area appears to be the site of ice production, and we consider the summer phase of the polynya to be \n348 \n‘closed’ (Table 1). To the east and north-east ice production is also taking place in the Pine Island Polynya, and ice-\n349 \npack that has drifted from the Bellingshausen Sea fills the western side of the Pine Island Polynya, adjacent to the \n350 \nAbbott Ice Shelf. 351 From April 2017 onwards, numerous polynya events take place. During these events ice in the polynya typ-\n352 \nically blows to the west, creating an opening in the polynya adjacent to all or part of the Iceberg Chain. New ice can \n353 10 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. be seen forming in this opening, typically as frazil bands, which sometimes form in approximately parallel bands \n354 \n(e.g. Fig. 2c-d). 4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 Otherwise, ice production sometimes forms a ‘swirl’ of thin ice. Alternatively, at times ice drifts \n355 \nnorthward from the Dotson Ice Shelf or Bear Peninsula and new ice forms adjacent to those sections while parts ad-\n356 \njacent to the Iceberg Chain remain closed. From April through August the polynya never extends further west than \n357 \nthe western margin of the Dotson Ice Shelf. However, separate ‘mini polynyas' within our ASP study area can be \n358 \nseen forming along the coast, particularly to the west, and sometimes north, from the rock outcrops between outlets \n359 \nof the Getz Ice Shelf. 360 Fig. 2. (a) An example image of the ASP during the winter in a Sentinel-1 SAR image from 5 September 2019. (b-\n361 \nd) An example of a polynya event taking place 21-23 September 2020 in Sentinel-1 SAR imagery. The area corre-\n362 \nsponds to the dashed-green box in (a). (e) The elevation of the bed referenced to mean sea level for the same area as \n363 \n(a). The bathymetry data is from the MEaSUREs BedMachine version 2 dataset (Morlighem et al., 2019). 364 \n \n365 Fig. 2. (a) An example image of the ASP during the winter in a Sentinel-1 SAR image from 5 September 2019. (b-\n361 \nd) An example of a polynya event taking place 21-23 September 2020 in Sentinel-1 SAR imagery. The area corre-\n362 \nsponds to the dashed-green box in (a). (e) The elevation of the bed referenced to mean sea level for the same area as \n363 \n(a). The bathymetry data is from the MEaSUREs BedMachine version 2 dataset (Morlighem et al., 2019). 364 \n365 Fig. 2. (a) An example image of the ASP during the winter in a Sentinel-1 SAR image from 5 September 2019. (b-\n361 \nd) An example of a polynya event taking place 21-23 September 2020 in Sentinel-1 SAR imagery. The area corre-\n362 \nsponds to the dashed-green box in (a). (e) The elevation of the bed referenced to mean sea level for the same area as \n363 \n(a). The bathymetry data is from the MEaSUREs BedMachine version 2 dataset (Morlighem et al., 2019). 364 \n365 11 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. From April through August 2017, new ice formed by the polynya extends and drifts westward overall, \n366 \neventually rounding the corner of Siple Island. 4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 A gap in the icepack also \n388 \nconnects the polynya to the open ocean in the north-west of our study area. Compared to the previous year, a more \n389 \nsubstantial section of icepack and fast ice separates the ASP from the Pine Island Polynya throughout the whole \n390 \nsummer. 391 \n \nBy late February the polynya is contained within the ASP study area, extending along the entire coast of \n392 \nthis area. There is also evidence of ice production in parts of the polynya around this time. By around 9 March 2018 \n393 \nice production is taking place across the polynya area (Table 1) and the open area gets progressively smaller through \n394 \nMarch and April, apart from during some polynya events. For example, a large event around 14 April takes place \n395 \nthat extends from the eastern boundary to the second-most eastern outlet of the Getz Ice Shelf. 396 \n \nThe polynya then behaves in a similar way to the previous year, with polynya events occurring throughout \n397 \nthe winter, and blockages of ice by the two grounded iceberg regions (next to Siple Island and the Central Grounded \n398 \nIcebergs) influencing the evacuation of new polynya-produced ice from the area. The ice that gets stuck in these ar-\n399 \neas is sometimes new polynya-produced ice and sometimes icepack that drifts in from the north-east. Some of this \n400 \nice appears to be formed in the Pine Island Polynya and some drifts in from the Bellingshausen Sea sector. When \n401 \nthis icepack comes into the polynya region, it appears to sometimes plays a role in pushing newly formed ice back \n402 4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 However, the ice does not consistently flow westward from the po-\n367 \nlynya. The ice often ‘heaves’ from the polynya, traveling westward overall but often reversing direction and tempo-\n368 \nrarily drifting back towards the polynya. When the polynya opens, it may then close due to new ice formation or be-\n369 \ncause polynya-produced ice drifts back into the open area. 370 \n \nThe flow of ice out of the study area is also disrupted by areas of ice that get stuck and prevent new ice \n371 \nfrom drifting westward along the coast. Between April and August an area of ice becomes stuck as it becomes fast to \n372 \nthe east of some grounded icebergs off the Siple Island coast. This forces ice to divert around as it rounds the corner \n373 \nof Siple Island. Ice also intermittently gets stuck/fast in a region around the center of the ASP study area we call the \n374 \n‘Central Grounded Icebergs’ (Fig 2a), which corresponds to a topographic high in the sea floor (Fig. 2e). The ob-\n375 \nstruction caused by this section of ice causes the polynya-produced ice to flow north around the Central Grounded \n376 \nIcebergs, leading to a substantial northward extension of polynya-produced ice, particularly in September-October. 377 \n \nFrom October larger polynya events begin to take place. For example, on 5 October and 25 October areas \n378 \nof open water occupied by newly-forming ice extend from the Iceberg Chain at least as far west as the eastern outlet \n379 \nof Getz Ice Shelf. In the latter part of November, the polynya opens and remains open, although ice production visi-\n380 \nbly takes place in some of the open area. By early December there is little evidence of ice production (Table 1) and \n381 \nthrough December the western boundary of the polynya progressively drifts to the west and north-west, extending \n382 \nthe polynya. 383 \n384 From April through August 2017, new ice formed by the polynya extends and drifts westward overall, \n366 \neventually rounding the corner of Siple Island. However, the ice does not consistently flow westward from the po-\n367 \nlynya. The ice often ‘heaves’ from the polynya, traveling westward overall but often reversing direction and tempo-\n368 \nrarily drifting back towards the polynya. When the polynya opens, it may then close due to new ice formation or be-\n369 \ncause polynya-produced ice drifts back into the open area. 4.1 Qualitative analysis of ASP using Sentinel-1 SAR imagery \n335 \n4.1.1 November 2016 - December 2017 \n336 370 From April through August 2017, new ice formed by the polynya extends and drifts westward overall, \n366 \neventually rounding the corner of Siple Island. However, the ice does not consistently flow westward from the po-\n367 \nlynya. The ice often ‘heaves’ from the polynya, traveling westward overall but often reversing direction and tempo-\n368 \nrarily drifting back towards the polynya. When the polynya opens, it may then close due to new ice formation or be-\n369 \ncause polynya-produced ice drifts back into the open area. 370 \n \nThe flow of ice out of the study area is also disrupted by areas of ice that get stuck and prevent new ice \n371 \nfrom drifting westward along the coast. Between April and August an area of ice becomes stuck as it becomes fast to \n372 \nthe east of some grounded icebergs off the Siple Island coast. This forces ice to divert around as it rounds the corner \n373 \nof Siple Island. Ice also intermittently gets stuck/fast in a region around the center of the ASP study area we call the \n374 \n‘Central Grounded Icebergs’ (Fig 2a), which corresponds to a topographic high in the sea floor (Fig. 2e). The ob-\n375 \nstruction caused by this section of ice causes the polynya-produced ice to flow north around the Central Grounded \n376 \nIcebergs, leading to a substantial northward extension of polynya-produced ice, particularly in September-October. 377 \n \nFrom October larger polynya events begin to take place. For example, on 5 October and 25 October areas \n378 \nof open water occupied by newly-forming ice extend from the Iceberg Chain at least as far west as the eastern outlet \n379 \nof Getz Ice Shelf. In the latter part of November, the polynya opens and remains open, although ice production visi-\n380 \nbly takes place in some of the open area. By early December there is little evidence of ice production (Table 1) and \n381 \nthrough December the western boundary of the polynya progressively drifts to the west and north-west, extending \n382 \nthe polynya. 383 \n \n384 \n4.1.2 January - December 2018 \n385 \n \nThrough the summer of 2017/18 (Video S1) the ocean to the north is occupied by more icepack than in \n386 \n2016/17 and therefore the polynya always has a northern boundary. However, through January the western boundary \n387 \nmigrates westward so that by early February it is outside our study area, if it remains at all. 4.1.2 January - December 2018 \n385 Through the summer of 2017/18 (Video S1) the ocean to the north is occupied by more icepack than in \n386 \n2016/17 and therefore the polynya always has a northern boundary. However, through January the western boundary \n387 \nmigrates westward so that by early February it is outside our study area, if it remains at all. A gap in the icepack also \n388 \nconnects the polynya to the open ocean in the north-west of our study area. Compared to the previous year, a more \n389 \nsubstantial section of icepack and fast ice separates the ASP from the Pine Island Polynya throughout the whole \n390 \nsummer. 391 By late February the polynya is contained within the ASP study area, extending along the entire coast of \n92 \nthis area. There is also evidence of ice production in parts of the polynya around this time. By around 9 March 2018 \n93 \nice production is taking place across the polynya area (Table 1) and the open area gets progressively smaller through \n94 \nMarch and April, apart from during some polynya events. For example, a large event around 14 April takes place \n95 \nthat extends from the eastern boundary to the second-most eastern outlet of the Getz Ice Shelf. 96 The polynya then behaves in a similar way to the previous year, with polynya events occurring throughout \n397 \nthe winter, and blockages of ice by the two grounded iceberg regions (next to Siple Island and the Central Grounded \n398 \nIcebergs) influencing the evacuation of new polynya-produced ice from the area. The ice that gets stuck in these ar-\n399 \neas is sometimes new polynya-produced ice and sometimes icepack that drifts in from the north-east. Some of this \n400 \nice appears to be formed in the Pine Island Polynya and some drifts in from the Bellingshausen Sea sector. When \n401 \nthis icepack comes into the polynya region, it appears to sometimes plays a role in pushing newly formed ice back \n402 12 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. into the open polynya when it ‘heaves’ and closes. By mid-November ice production appears to mostly cease and \n403 \nthe polynya opens up further through November and December (Table 1). 4.1.2 January - December 2018 \n385 From then the polynya behaves in a similar manner to the previous \n413 \ntwo years. 414 As in the previous year, a more substantial icepack than in 2016/17 exists to the north of the ASP through \n407 \nthe summer of 2018/19. However, in the summer of 2018/19 this icepack connects with the coast, giving the ASP a \n408 \nwestern limit and keeping the ASP isolated from the open ocean throughout the summer. Compared to the previous \n409 \nyear, the neighboring Pine Island Polynya extends closer to the ASP but a section of ice fast along the eastern side of \n410 \nthe Iceberg Chain maintains a clear boundary between the two polynyas. Through February the icepack flows \n411 \naround the Iceberg Chain and in from the north, reducing the polynya’s area. By 14 March 2019 new ice or forming \n412 \nice is visible across the whole polynya (Table 1). From then the polynya behaves in a similar manner to the previous \n413 \ntwo years. 414 Year \nPolynya opens \nPolynya closes \n2016/17 \n8 November 2016* \n4 April 2017 \n2017/18 \n3 December 2017 \n9 March 2018 \n2018/19 \n13 November 2018 \n14 March 2019 \n2019/20 \n20 November 2019 \n21 March 2020 \n2020/21 \n16 November 2020 \n8 March 2021 415 \nTable 1. Summer polynya opening and closing dates for each summer 2016/17-2020/21 as determined by visual \n416 \nanalysis of Sentinel-1 SAR imagery. We determine the polynya to be open for summer when the majority of the \n417 \nopen polynya is not exhibiting ice production and closed when the majority of the polynya is exhibiting ice produc-\n418 \ntion. * in 2016/17 a lack of imagery in early November means it is difficult to determine when the polynya opened, \n419 \nbut it is open by 8 November. 420 \n \n421 \nOne notable difference in the winter of 2019 compared to the previous years is that, until October, newly \n422 \npolynya-produced ice flows westward only on the southern (coastal) side of the Central Grounded Icebergs. Also, \n423 \nthrough August, and most of September, icepack settles immediately adjacent to the northern section of the Iceberg \n424 \nChain. This area is otherwise typically occupied by open polynya or polynya-produced ice. Both of these factors \n425 \nmean that the polynya-produced ice forms a narrower band closer to the coast. 4.1.2 January - December 2018 \n385 404 \n \n405 \n4.1.3 January - December 2019 \n406 \n \nAs in the previous year, a more substantial icepack than in 2016/17 exists to the north of the ASP through \n407 \nthe summer of 2018/19. However, in the summer of 2018/19 this icepack connects with the coast, giving the ASP a \n408 \nwestern limit and keeping the ASP isolated from the open ocean throughout the summer. Compared to the previous \n409 \nyear, the neighboring Pine Island Polynya extends closer to the ASP but a section of ice fast along the eastern side of \n410 \nthe Iceberg Chain maintains a clear boundary between the two polynyas. Through February the icepack flows \n411 \naround the Iceberg Chain and in from the north, reducing the polynya’s area. By 14 March 2019 new ice or forming \n412 \nice is visible across the whole polynya (Table 1). From then the polynya behaves in a similar manner to the previous \n413 \ntwo years. 414 into the open polynya when it ‘heaves’ and closes. By mid-November ice production appears to mostly cease and \n403 \nthe polynya opens up further through November and December (Table 1). 404 \n405 into the open polynya when it ‘heaves’ and closes. By mid-November ice production appears to mostly cease and \n403 \nthe polynya opens up further through November and December (Table 1). 404 into the open polynya when it ‘heaves’ and closes. By mid-November ice production appears to mostly cease and \n403 \nthe polynya opens up further through November and December (Table 1). 404 \n \n405 \n4.1.3 January - December 2019 \n406 \n \nAs in the previous year, a more substantial icepack than in 2016/17 exists to the north of the ASP through \n407 \nthe summer of 2018/19. However, in the summer of 2018/19 this icepack connects with the coast, giving the ASP a \n408 \nwestern limit and keeping the ASP isolated from the open ocean throughout the summer. Compared to the previous \n409 \nyear, the neighboring Pine Island Polynya extends closer to the ASP but a section of ice fast along the eastern side of \n410 \nthe Iceberg Chain maintains a clear boundary between the two polynyas. Through February the icepack flows \n411 \naround the Iceberg Chain and in from the north, reducing the polynya’s area. By 14 March 2019 new ice or forming \n412 \nice is visible across the whole polynya (Table 1). 4.1.2 January - December 2018 \n385 This is the case until October when \n426 \nice becomes stuck to the south of the Central Grounded Icebergs causing the polynya-produced ice to divert around \n427 \nthe northern side of the Central Grounded Icebergs. 428 \n \nIn October and early November there are persistent polynya events and the new polynya-produced ice con-\n429 \nsistently moves westward, away from the polynya. By around 20 November there is an open polynya area without \n430 In October and early November there are persistent polynya events and the new polynya-produced ice con-\n429 \nsistently moves westward, away from the polynya. By around 20 November there is an open polynya area without \n430 13 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. substantial ice production across most of its area (Table 1). The polynya then extends to the west through November \n431 \nand December. 432 stantial ice production across most of its area (Table 1). The polynya then extends to the west through November \nd December. substantial ice production across most of its area (Table 1). The polynya then extends to the west through November \n431 \nd D\nb\n432 substantial ice production across most of its area (Table 1). The polynya then extends to the west through November \n431 \n432 Fig. 3. Ice inside the red boundary is estimated, based on visual analysis of Video S1, to be approximately all the ice \n433 \nproduced by the main polynya between 30 April and 4 November 2020. Sentinel-1 SAR image from 4 November \n434 \n2020. 435 \n \n436 \n4.1.3 January 2020 - March 2021 \n437 \n \nAs in the summer of 2018/19, throughout the summer of 2019/20 the polynya remains completely bound by \n438 \nicepack and isolated from the open ocean (Video S1). A more substantial area of icepack and fast ice lies between \n439 \nthe ASP and the Pine Island Polynya, similar to the summer of 2017/18. Ice production is visibly taking place by \n440 \nearly March and by 20 March 2020 new ice/ice-production is visible across the whole polynya (Table 1). Through \n441 Fig. 3. Ice inside the red boundary is estimated, based on visual analysis of Video S1, to be approximately all the ice \n433 \nproduced by the main polynya between 30 April and 4 November 2020. Sentinel-1 SAR image from 4 November \n434 \n2020. 4.1.2 January - December 2018 \n385 435 As in the summer of 2018/19, throughout the summer of 2019/20 the polynya remains completely bound by \n438 \nicepack and isolated from the open ocean (Video S1). A more substantial area of icepack and fast ice lies between \n439 \nthe ASP and the Pine Island Polynya, similar to the summer of 2017/18. Ice production is visibly taking place by \n440 \nearly March and by 20 March 2020 new ice/ice-production is visible across the whole polynya (Table 1). Through \n441 14 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. April-June, as in other years, regular polynya events originate from the Iceberg Chain and the Dotson Ice Shelf. In \n442 \nlate June icepack rounds the Iceberg Chain and pushes southward into the ASP, as happened in August-September \n443 \nof 2019. There then follows a period largely lacking in imagery, but visual analysis of the polynya-produced ice in \n444 \nthe region suggests that there is little ice production from then until early August. At this time the polynya-produced \n445 \nice that formed prior to 30 April has become stuck adjacent to the Central Grounded Icebergs, south and south-east \n446 \nof the icebergs. This causes the new ice produced in the ASP from early August onwards to divert westward on the \n447 \nnorth side of the Central Grounded Icebergs, entraining some fragments of the stuck, pre-August ice. From August \n448 \nonwards, after some stuck ice becomes dislodged, a separate small ‘secondary polynya (Fig. 2a) also repeatedly \n449 \nforms on the west side of the Central Iceberg Chain. New ice created by this small secondary polynya mixes with ice \n450 \nformed by the main ASP and flows westward around Siple Island. Through November the polynya-produced ice \n451 \ndrifts away from the Iceberg Chain and Dotson Ice Shelf, and by 16 November ice production appears to have \n452 \nmostly ceased (Table 1) as it enters the summer period. The summer polynya closes around 8 March 2021. 453 By analyzing Video S1 it is sometimes possible to visually track ice produced by particular polynya events \n454 \nthrough the season. In particular, in 2020, with some uncertainty due to missing images, we are able to estimate that \n455 \napproximately all of the ice produced between 30 April and 4 November by the main polynya (i.e. 4.1.2 January - December 2018 \n385 In 2017/18, 2019/20 and 2020/21 the polynya behaves in a similar \n477 \nmanner through most the period, with no one of those years consistently recording a higher area, and each year \n478 \nreaching a peak-open area that approximately fills the whole ASP study area. However, polynya area in 2020/21 \n479 \nreaches its peak later (January), and its decline begins later (late February). Notably, in 2017/18 the polynya experi-\n480 \nences a temporary rapid re-opening as it increases from just 5 977 km2 on 15 March to 48 7779 km2 on 21 March. 481 \nThis is an 82x increase in 6 days, and it is followed by a rapid decline. The polynya had the highest daily mean area \n482 \nfor summer (November-March) in 2016/17, at 62 616 km2, and 2018/19 had the lowest, at 38518 km2. The mean \n483 \ndaily area of 2017/18, 2019/20 and 2020/21 for summer was 44 013 km2, 44 979 km2 and 44 447 km2, respectively. 484 4.2 Summer Polynya Area \n463 \n \nIn all years there is an overall increase in polynya area through November (Fig 4). On 1 November, the po-\n464 \nlynya has an area between 17 813 km2 (2019) and 25 859 km2 (2016). In the years 2017/18, 2018/19, 2019/20 and \n465 \n2020/21 the polynya area then follows a similar pattern, but in 2016/17 it follows a distinct course. By 1 December \n466 \nin 2016 the polynya is open in approximately the whole ASP study area, with an area of 65 674 km2, an increase of \n467 \n154% from 1 November. In 2016/17 the polynya remains open across approximately the whole study area through-\n468 \nout December, January, and most of February and March, only beginning to significantly decline in late March (Fig \n469 \n4) and into April (Fig 5). The polynya in 2016/17 maintains a higher area than in all other years throughout the \n470 \nwhole summer, apart from a small period in late February when it is surpassed by 2020/21. 471 In 2017/18, 2018/19, 2020/21 and 2020/21 the polynya area has increased to between 25 439 km2 (2017) \n472 \nand 38 310 km2 (2020) by 1 December. From then the polynya continues to follow an overall increasing trend \n473 \nthrough December, with the polynya reaching its peak area in January in each of these years. 4.1.2 January - December 2018 \n385 excluding the \n456 \nsecondary polynya located by the Central Grounded Icebergs) is contained within the red outline on 4 November in \n457 \nFig. 3. This section of polynya-produced ice totals 46452 km2 in area. 458 Fig. 4. Daily summer (November-March) polynya area for each summer 2016/17-2020/21 (solid), and the daily \n459 \nmean for the whole period (dashed). 460 \n461 Fig. 4. Daily summer (November-March) polynya area for each summer 2016/17-2020/21 (solid), and the daily \n459 \nmean for the whole period (dashed). 460 \n461 15 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 4.2 Summer Polynya Area \n463 \n \nIn all years there is an overall increase in polynya area through November (Fig 4). On 1 November, the po-\n464 \nlynya has an area between 17 813 km2 (2019) and 25 859 km2 (2016). In the years 2017/18, 2018/19, 2019/20 and \n465 \n2020/21 the polynya area then follows a similar pattern, but in 2016/17 it follows a distinct course. By 1 December \n466 \nin 2016 the polynya is open in approximately the whole ASP study area, with an area of 65 674 km2, an increase of \n467 \n154% from 1 November. In 2016/17 the polynya remains open across approximately the whole study area through-\n468 \nout December, January, and most of February and March, only beginning to significantly decline in late March (Fig \n469 \n4) and into April (Fig 5). The polynya in 2016/17 maintains a higher area than in all other years throughout the \n470 \nwhole summer, apart from a small period in late February when it is surpassed by 2020/21. 471 \n \nIn 2017/18, 2018/19, 2020/21 and 2020/21 the polynya area has increased to between 25 439 km2 (2017) \n472 \nand 38 310 km2 (2020) by 1 December. From then the polynya continues to follow an overall increasing trend \n473 \nthrough December, with the polynya reaching its peak area in January in each of these years. In 2018/19 the peak \n474 \narea is substantially lower (61 113 km2) than in other years, and the polynya only maintains an area above 60 000 \n475 \nkm2 for six days in late January and early February. 2018/19 records the lowest area in comparison to other years on \n476 \nevery day between 4 December and 3 February. 4.1.2 January - December 2018 \n385 In 2018/19 the peak \n474 \narea is substantially lower (61 113 km2) than in other years, and the polynya only maintains an area above 60 000 \n475 \nkm2 for six days in late January and early February. 2018/19 records the lowest area in comparison to other years on \n476 \nevery day between 4 December and 3 February. In 2017/18, 2019/20 and 2020/21 the polynya behaves in a similar \n477 \nmanner through most the period, with no one of those years consistently recording a higher area, and each year \n478 \nreaching a peak-open area that approximately fills the whole ASP study area. However, polynya area in 2020/21 \n479 \nreaches its peak later (January), and its decline begins later (late February). Notably, in 2017/18 the polynya experi-\n480 \nences a temporary rapid re-opening as it increases from just 5 977 km2 on 15 March to 48 7779 km2 on 21 March. 481 \nThis is an 82x increase in 6 days, and it is followed by a rapid decline. The polynya had the highest daily mean area \n482 \nfor summer (November-March) in 2016/17, at 62 616 km2, and 2018/19 had the lowest, at 38518 km2. The mean \n483 \ndaily area of 2017/18, 2019/20 and 2020/21 for summer was 44 013 km2, 44 979 km2 and 44 447 km2, respectively. 484 16 16 Fig. 5. Daily winter (April-October) polynya area for each winter 2017-2020 (solid), and the daily mean for the \n485 \nwhole period, as measured from AMSR-2 SIC data (dashed). 486 \n487 Fig. 5. Daily winter (April-October) polynya area for each winter 2017-2020 (solid), and the daily mean for the \n485 \nwhole period, as measured from AMSR-2 SIC data (dashed). 486 \n487 4.3 Winter Polynya Area and Ice Production \n488 507 \n \nAnalysis of the spatial distribution of ice production across all years reveals that the mean daily ice produc-\n508 \ntion is highest in the area of the polynya adjacent to the Iceberg Chain, Thwaites Iceberg Tongue and Dotson Ice \n509 \nShelf (Fig. 7). Mean annual ice production values (April-October) in this region surpass 17 m3/m2. Other notable \n510 \nareas of higher ice production lie along various parts of the coast and an area that corresponds to the secondary po-\n511 \nlynya by the Central Grounded Icebergs. 512 We note that in 2017 between the dates of 1 April and 8 May a substantial portion of the calculated open \n503 \npolynya area, and ice production, occurs in the northwest of the ASP study area. This area is part of the open ocean \n504 \nand is separated by sea ice from the more-typically open polynya area adjacent to the iceberg chain. Typically ice-\n505 \npack fills this northwest area, but in early 2017 it is open due to the lack of icepack in this sector in the summer of \n506 \n2016/17, discussed in section 4.3 (Video S2; Fig. 11b). 507 \n \nAnalysis of the spatial distribution of ice production across all years reveals that the mean daily ice produc-\n508 \ntion is highest in the area of the polynya adjacent to the Iceberg Chain, Thwaites Iceberg Tongue and Dotson Ice \n509 \nShelf (Fig. 7). Mean annual ice production values (April-October) in this region surpass 17 m3/m2. Other notable \n510 \nareas of higher ice production lie along various parts of the coast and an area that corresponds to the secondary po-\n511 \nlynya by the Central Grounded Icebergs. 512 We note that in 2017 between the dates of 1 April and 8 May a substantial portion of the calculated open \n503 \npolynya area, and ice production, occurs in the northwest of the ASP study area. This area is part of the open ocean \n504 \nand is separated by sea ice from the more-typically open polynya area adjacent to the iceberg chain. Typically ice-\n505 \npack fills this northwest area, but in early 2017 it is open due to the lack of icepack in this sector in the summer of \n506 \n2016/17, discussed in section 4.3 (Video S2; Fig. 11b). 4.3 Winter Polynya Area and Ice Production \n488 In all years, polynya area exhibits an overall decline from the beginning of the winter period, when the po-\n489 \nlynya remains relatively large after the summer period (Fig. 5). This period, when the polynya remains relatively \n490 \nlarge but ice production has now begun, accounts for a substantial proportion of the annual ice production (Fig. 6; \n491 \nFig. S2). On average, April/May accounts for 36% (39.6 km3) of annual ice production. The polynya then generally \n492 \nreaches a sustained winter low in area, where it fluctuates around and below 10 000 km2. In 2020 the polynya area \n493 \nreaches its low in early April, while in 2017, 2019 and mean 2017-20 the area continues an overall decline through \n494 \nApril, May and June. Polynya area and ice production then tend to remain low until an increase begins around Sep-\n495 \ntember. In July polynya area remains below 10 000 km2 in all years for the whole month apart from brief small fluc-\n496 \ntuations above this in 2018 and 2019. There are notable spikes in polynya area in the middle of the year, which are \n497 \nalso exhibited in spikes in ice production. Most notably in June in 2018 polynya area spikes to 26 631 km2. After a \n498 \nperiod of low polynya area, the area generally increases through September and October towards the summer period. 499 \nIn 2020 this period of area and ice production increase begins in August. This late-winter increase in polynya area is \n500 \nalso exhibited in a corresponding marked increase in the rate of ice production. On average, September/October ac-\n501 \ncounts for 42% (45.8 km3) of annual ice production. 502 17 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. We note that in 2017 between the dates of 1 April and 8 May a substantial portion of the calculated open \n503 \npolynya area, and ice production, occurs in the northwest of the ASP study area. This area is part of the open ocean \n504 \nand is separated by sea ice from the more-typically open polynya area adjacent to the iceberg chain. Typically ice-\n505 \npack fills this northwest area, but in early 2017 it is open due to the lack of icepack in this sector in the summer of \n506 \n2016/17, discussed in section 4.3 (Video S2; Fig. 11b). 4.3 Winter Polynya Area and Ice Production \n488 507 \n \nAnalysis of the spatial distribution of ice production across all years reveals that the mean daily ice produc-\n508 \ntion is highest in the area of the polynya adjacent to the Iceberg Chain, Thwaites Iceberg Tongue and Dotson Ice \n509 \nShelf (Fig. 7). Mean annual ice production values (April-October) in this region surpass 17 m3/m2. Other notable \n510 \nareas of higher ice production lie along various parts of the coast and an area that corresponds to the secondary po-\n511 \nlynya by the Central Grounded Icebergs. 512 Fig. 6. Daily cumulative winter ice production for each winter (April-October) 2017-2020 (solid), and the mean for \n513 \nthe period (dashed), as measured using heat-flux modeling of ERA-5 data and AMSR-2 SIC data. Also shown are \n514 \nmean annual measurements for 1992-2001 (Tamura et al., 2008), 1992-2013 (Tamura et al., 2016), 2003-10 and \n515 \n2013-15 (Nihashi et al., 2017), along with the instrument used for each measurement. Note the previous studies’ \n516 \nmeasurements covered the period March-October and used a study area that does not exactly correspond to ours. 517 \n \n518 Fig. 6. Daily cumulative winter ice production for each winter (April-October) 2017-2020 (solid), and \n513 Fig. 6. Daily cumulative winter ice production for each winter (April-October) 2017-2020 (solid), and the mean for \n513 \nthe period (dashed), as measured using heat-flux modeling of ERA-5 data and AMSR-2 SIC data. Also shown are \n514 \nmean annual measurements for 1992-2001 (Tamura et al., 2008), 1992-2013 (Tamura et al., 2016), 2003-10 and \n515 \n2013-15 (Nihashi et al., 2017), along with the instrument used for each measurement. Note the previous studies’ \n516 \nmeasurements covered the period March-October and used a study area that does not exactly correspond to ours. 517 \n \n518 18 18 Table 2. Estimates of Mean Daily Polynya Area, Total Annual Ice Production and Mean Daily Ice Production dur-\ning the winters of 2017-2020, and for the daily mean of the period. Numbers in brackets indicate the standard devia-\ntion. Year \nMean Daily Polynya \nArea (km2 ) \nTotal Annual Ice \nProduction (km3 ) \nTotal Mean Daily Ice \nProduction (km3 ) \n2017 \n10 908 (9589) \n139 \n0.67 (0.67) \n2018 \n9 963 (7004) \n121 \n0.57 (0.50) \n2019 \n8 152 (5127) \n95 \n0.45 (0.38) \n2020 \n6 910 (5692) \n80 \n0.38 (0.36) \nMean 2017-20 \n8 984 (7240) \n109 \n0.52 (0.51) Table 2. 4.3 Winter Polynya Area and Ice Production \n488 Many day-to-day variations in polynya area are not correlated with \n522 \nwind speed, however it is clear that notable spikes in polynya area do often occur on days with high wind speed \n523 \n(Fig. S3). For example, the three highest polynya areas recorded in 2020 after April all occur alongside \n524 \n the three highest spikes in wind speed post-April. By viewing the mean spatial distribution of wind it is clear that \n525 \nthe ASP forms in an area of relatively high winds (Fig. 9). A band of high winds with a mean speed of around 8 - 9 \n526 \nms-1 exists along the coast from Thwaites Glacier, over the Thwaites Iceberg Tongue and into the eastern area of the \n527 \nASP study area, where the main polynya originates. 528 The mean wind direction throughout the ASP study area is approximately southerly. While this direction \n529 \ncorresponds to the direction in which the polynya sometimes forms northward off the Dotson Ice Shelf, it does not \n530 \ncorrespond to the more typical westward formation off the iceberg chain. 531 Fig. 8. Scatter plot of daily polynya area and wind speed for all winter days (April-October 2017-2020), except for \n532 \nthe period 1 April - 8 May 2017 (which is excluded because during this time the study site included a substantial \n533 \nopen area on the seaward side of icepack and at a distance from the area used to calculate wind speed (Fig. 11b)). 534 \nThe Pearson Product Correlation Coefficient is 0.33 (P < 0.05). 535 \n \n536 Fig. 8. Scatter plot of daily polynya area and wind speed for all winter days (April-October 2017-2020), except for \n532 \nthe period 1 April - 8 May 2017 (which is excluded because during this time the study site included a substantial \n533 \nopen area on the seaward side of icepack and at a distance from the area used to calculate wind speed (Fig. 11b)). 534 \nThe Pearson Product Correlation Coefficient is 0.33 (P < 0.05). 535 \n \n536 \n \n537 \n \n538 \n \n539 Fig. 8. Scatter plot of daily polynya area and wind speed for all winter days (April-October 2017-2020), except for \n532 \nthe period 1 April - 8 May 2017 (which is excluded because during this time the study site included a substantial \n533 \nopen area on the seaward side of icepack and at a distance from the area used to calculate wind speed (Fig. 11b)). 4.3 Winter Polynya Area and Ice Production \n488 Estimates of Mean Daily Polynya Area, Total Annual Ice Production and Mean Daily Ice Production dur-\ning the winters of 2017-2020, and for the daily mean of the period. Numbers in brackets indicate the standard devia-\ntion. Table 2. Estimates of Mean Daily Polynya Area, Total Annual Ice Production and Mean Daily Ice Production dur-\ning the winters of 2017-2020, and for the daily mean of the period. Numbers in brackets indicate the standard devia-\ntion. Fig. 7. Mean Annual Ice Production for all winter study periods (April-October 2017-2020). The region corresponds \nto the region within the green outline in Fig. 1b Fig. 7. Mean Annual Ice Production for all winter study periods (April-October 2017-2020). The region corresponds \nto the region within the green outline in Fig. 1b 519 19 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 4.4 Wind and polynya area \n520 \n \nDaily mean wind speed at the polynya site and polynya area during the winter period has a weak but signif-\n521 \nicant positive correlation (0.33, P <0.05) (Fig. 8). Many day-to-day variations in polynya area are not correlated with \n522 \nwind speed, however it is clear that notable spikes in polynya area do often occur on days with high wind speed \n523 \n(Fig. S3). For example, the three highest polynya areas recorded in 2020 after April all occur alongside \n524 \n the three highest spikes in wind speed post-April. By viewing the mean spatial distribution of wind it is clear that \n525 \nthe ASP forms in an area of relatively high winds (Fig. 9). A band of high winds with a mean speed of around 8 - 9 \n526 \nms-1 exists along the coast from Thwaites Glacier, over the Thwaites Iceberg Tongue and into the eastern area of the \n527 \nASP study area, where the main polynya originates. 528 \n \nThe mean wind direction throughout the ASP study area is approximately southerly. While this direction \n529 \ncorresponds to the direction in which the polynya sometimes forms northward off the Dotson Ice Shelf, it does not \n530 \ncorrespond to the more typical westward formation off the iceberg chain. 531 4.4 Wind and polynya area \n520 \n \nDaily mean wind speed at the polynya site and polynya area during the winter period has a weak but signif-\n521 \nicant positive correlation (0.33, P <0.05) (Fig. 8). 4.3 Winter Polynya Area and Ice Production \n488 534 \nThe Pearson Product Correlation Coefficient is 0.33 (P < 0.05). 535 \n536 20 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. g. 9. Mean wind speed and direction for the period 1 November 2016 to 31 December 2020. The green box repre-\nnts the ASP study area, the same as the green box in Fig 1b. Daily wind speed and direction is included as Video \n3. Mean wind speed and direction for the period 1 November 2016 to 31 December 2020. The green box repre-\nhe ASP study area, the same as the green box in Fig 1b. Daily wind speed and direction is included as Video Fig. 9. Mean wind speed and direction for the period 1 November 2016 to 31 December 2020. The green box repre-\n540 \nsents the ASP study area, the same as the green box in Fig 1b. Daily wind speed and direction is included as Video \n541 Fig. 9. Mean wind speed and direction for the period 1 November 2016 to 31 December 2020. The green box repre-\n540 \nsents the ASP study area, the same as the green box in Fig 1b. Daily wind speed and direction is included as Video \n541 \nS3. 542 Fig. 9. Mean wind speed and direction for the period 1 November 2016 to 31 December 2020. The green box repre-\n540 \nsents the ASP study area, the same as the green box in Fig 1b. Daily wind speed and direction is included as Video \n541 \nS3. 542 \n543 21 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. 22 \n \n547 \n \n548 \n \n549 \n \n550 \n \n551 \n \n552 \n \n553 \n \n554 \n \n555 \n \n556 \n \n557 \n \n558 \n \n559 \n \n560 \n \n561 547 22 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. The icepack continues to diminish and the gap con-\n587 \nnecting the ASP to the open ocean broadens until by February the polynya is only bound by part of the Iceberg \n588 \nChain and Thwaites Iceberg Tongue. The total SIC reaches a minimum of 421 434 on 5 February 2017, 35% of the \n589 \nnext lowest annual minimum (2019/20). Gaps in icepack around the Iceberg Chain mean that the ASP has essen-\n590 \ntially joined with the Pine Island Polynya through the Iceberg Chain for a period in this year. The narrow band of \n591 Fig. 11. SIC for the broader ASP region on two days in 2017, during and following a summer of record-low SIC. 575 \nThe area corresponds to that shown by red box in Fig. 1a. 576 \n \n577 Fig. 11. SIC for the broader ASP region on two days in 2017, during and following a summer of record-low SIC. 575 \nThe area corresponds to that shown by red box in Fig. 1a. 576 4.5 Broader SIC Analysis of the SIC over a broader area also shows daily changes in the polynya area and how it relates to \n579 \nchanges in the icepack. The mean monthly cycle of the polynya can be seen in Fig. 10 and presents a similar picture \n580 \nof the polynya as described in sections 4.1- 4.3. The broader icepack has a minimum total SIC of 1 975 921 in Janu-\n581 \nary and remains similar in February. From March the broader icepack can be seen to expand in area as the polynya \n582 \nbegins to close and continues to increase until a peak of 6 588 615 cumulative SIC in September. From October the \n583 \nicepack begins a marked decline into summer. Interannual variation can be seen in Fig. S4 and Video S2, with maxi-\n584 \nmum icepack area occurring in August or September each year, and minimum icepack in January or February. 585 Analysis of the SIC over a broader area also shows daily changes in the polynya area and how it relates to \n579 \nchanges in the icepack. The mean monthly cycle of the polynya can be seen in Fig. 10 and presents a similar picture \n580 \nof the polynya as described in sections 4.1- 4.3. The broader icepack has a minimum total SIC of 1 975 921 in Janu-\n581 \nary and remains similar in February. From March the broader icepack can be seen to expand in area as the polynya \n582 \nbegins to close and continues to increase until a peak of 6 588 615 cumulative SIC in September. From October the \n583 \nicepack begins a marked decline into summer. Interannual variation can be seen in Fig. S4 and Video S2, with maxi-\n584 \nmum icepack area occurring in August or September each year, and minimum icepack in January or February. 585 \n \nDuring the summer of 2016/17 the icepack is notably sparse (Fig. 11; S4). Around 29 December 2016 a \n586 \ngap in the icepack connects the ASP to the open ocean to north. The icepack continues to diminish and the gap con-\n587 \nnecting the ASP to the open ocean broadens until by February the polynya is only bound by part of the Iceberg \n588 \nChain and Thwaites Iceberg Tongue. The total SIC reaches a minimum of 421 434 on 5 February 2017, 35% of the \n589 \nnext lowest annual minimum (2019/20). https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Fig. 10. Mean monthly SIC for the broader ASP region for the period November 2016 to March 2021. The area cor-\n562 \nresponds to that shown by red box in Fig. 1a. Daily data is shown in Video S2. 563 \n564 Fig. 10. Mean monthly SIC for the broader ASP region for the period November 2016 to March 2021. The area cor-\n562 \nresponds to that shown by red box in Fig. 1a. Daily data is shown in Video S2. 563 Fig. 10. Mean monthly SIC for the broader ASP region for the period November 2016 to March 2021. The area cor-\n562 \nresponds to that shown by red box in Fig. 1a. Daily data is shown in Video S2. 563 23 Fig. 11. SIC for the broader ASP region on two days in 2017, during and following a summer of record-low SIC. 575 \nThe area corresponds to that shown by red box in Fig. 1a. 576 \n577 Fig. 11. SIC for the broader ASP region on two days in 2017, during and following a summer of record-low SIC. 575 \nThe area corresponds to that shown by red box in Fig. 1a. 576 \n \n577 \n4.5 Broader SIC \n578 \n \nAnalysis of the SIC over a broader area also shows daily changes in the polynya area and how it relates to \n579 \nchanges in the icepack. The mean monthly cycle of the polynya can be seen in Fig. 10 and presents a similar picture \n580 \nof the polynya as described in sections 4.1- 4.3. The broader icepack has a minimum total SIC of 1 975 921 in Janu-\n581 \nary and remains similar in February. From March the broader icepack can be seen to expand in area as the polynya \n582 \nbegins to close and continues to increase until a peak of 6 588 615 cumulative SIC in September. From October the \n583 \nicepack begins a marked decline into summer. Interannual variation can be seen in Fig. S4 and Video S2, with maxi-\n584 \nmum icepack area occurring in August or September each year, and minimum icepack in January or February. 585 \n \nDuring the summer of 2016/17 the icepack is notably sparse (Fig. 11; S4). Around 29 December 2016 a \n586 \ngap in the icepack connects the ASP to the open ocean to north. 4.5 Broader SIC 4, \n597 \nVideo S1), while during the winter it becomes smaller, opening up during ice-producing polynya ‘events’ (Figs. 5-6, \n598 \nVideo S1). These polynya events and changes in polynya area can sometimes be attributed to higher wind speeds \n599 \n(Figs 8-9, S3). 600 Our qualitative analysis of Sentinel-1 SAR imagery, however, also reveals distinct characteristics of the \n601 \nASP which are not possible to decipher from sea ice concentration data, other quantitative methods, or indirect ob-\n602 \nservation. First, we note that while in many other polynyas, such as the Ross Ice Shelf Polynya, new polynya-pro-\n603 \nduced ice is typically efficiently evacuated away from its origin (Dai et al., 2020), this is not the case at the ASP. 604 \nInstead, ice formed by the ASP often remains in the ASP study area for months (Video S1, Fig. 3). In fact, this po-\n605 \nlynya-produced ice does not consistently flow in a direction away from the polynya. While its overall direction is \n606 \nwestward, away from the polynya, the ice ‘heaves’ and temporarily flows ‘backward’ (‘back-fills’), as has also been \n607 \nobserved at the Mertz Glacier Polynya (Massom et al., 2017). The ice also gets ‘stuck’ in the region, particularly \n608 \naround grounded icebergs, which are grounded in areas where there are topographic highs in the sea floor (Fig. 2a, \n609 \ne). 610 The tendency of ice produced by the ASP to remain in the vicinity of the polynya for prolonged periods, \n611 \nbecome stuck, and sometimes flow back eastward towards the site of its formation, influences the polynya and ice \n612 \nproduction in two key ways. First, when ice moves eastward while the polynya is open it contributes to the closure \n613 \nof the polynya, and thus the cessation of ice production. Typically it is assumed that an open polynya during winter \n614 \ncloses due to new ice production (e.g. Cheng et al., 2017). However, the ASP may close both due to ice production \n615 \nand movement of previously-produced ice into the polynya. Second, we suggest that the blockages of ice in the vi-\n616 \ncinity of the polynya reduce the size and frequency of polynya events by hindering the ability of ice to move out of \n617 \nthe polynya and open it up. Again, by reducing the size and duration of open polynya area during winter, ice produc-\n618 \ntion is limited. 4.5 Broader SIC First, when ice moves eastward while the polynya is open it contributes to the closure \n613 \nof the polynya, and thus the cessation of ice production. Typically it is assumed that an open polynya during winter \n614 \ncloses due to new ice production (e.g. Cheng et al., 2017). However, the ASP may close both due to ice production \n615 \nand movement of previously-produced ice into the polynya. Second, we suggest that the blockages of ice in the vi-\n616 \ncinity of the polynya reduce the size and frequency of polynya events by hindering the ability of ice to move out of \n617 \nthe polynya and open it up. Again, by reducing the size and duration of open polynya area during winter, ice produc-\n618 \ntion is limited. 619 \n \nWe also note that the ASP forms along the coast westward off a chain of icebergs that extend from the \n620 \nThwaites Iceberg Tongue. While some polynyas, such as the Ross Ice Shelf polynya, form off and away from the \n621 \ncoast, the majority of Antarctic polynyas have been shown to form westward off glacier ice tongues or protruding \n622 \nfast ice (Nihashi et al., 2017). In the case of the ASP, its location and orientation is determined by the presence of \n623 \nthe ‘Iceberg Chain’, which is in turn determined by the presence of a bathymetric high (Figs. 1-2). Stammerjohn et \n624 \nal. (2015) refer to the polynya as forming off a ‘fast ice tongue’ but we prefer to refer to the ‘Iceberg Chain’ as the \n625 \neastern boundary. While a section of fast ice exists amongst, and adjacent to, a section of the southern part of the \n626 \nIceberg Chain, the extent of this fast ice varies and it only ever extends along a portion of the Iceberg Chain. The \n627 \nIceberg Chain remains virtually the same length throughout our observations. The polynya consistently forms off the \n628 icepack from the east closes the polynya off again in March but the band of adjacent icepack remains so narrow that, \n592 \nas mentioned, there is open ocean inside our ASP study area until 9 May (Video S2). 593 \n594 Our analysis shows that in some ways the ASP, between November 2016 and March 2021, behaves as is \n596 \ntypical for Antarctic coastal polynyas. During the summer the polynya becomes larger and remains ice-free (Fig. 4.5 Broader SIC Gaps in icepack around the Iceberg Chain mean that the ASP has essen-\n590 \ntially joined with the Pine Island Polynya through the Iceberg Chain for a period in this year. The narrow band of \n591 24 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. icepack from the east closes the polynya off again in March but the band of adjacent icepack remains so narrow that, \n592 \nas mentioned, there is open ocean inside our ASP study area until 9 May (Video S2). 593 \n \n594 \n5. Discussion \n \n595 \n \nOur analysis shows that in some ways the ASP, between November 2016 and March 2021, behaves as is \n596 \ntypical for Antarctic coastal polynyas. During the summer the polynya becomes larger and remains ice-free (Fig. 4, \n597 \nVideo S1), while during the winter it becomes smaller, opening up during ice-producing polynya ‘events’ (Figs. 5-6, \n598 \nVideo S1). These polynya events and changes in polynya area can sometimes be attributed to higher wind speeds \n599 \n(Figs 8-9, S3). 600 \n \nOur qualitative analysis of Sentinel-1 SAR imagery, however, also reveals distinct characteristics of the \n601 \nASP which are not possible to decipher from sea ice concentration data, other quantitative methods, or indirect ob-\n602 \nservation. First, we note that while in many other polynyas, such as the Ross Ice Shelf Polynya, new polynya-pro-\n603 \nduced ice is typically efficiently evacuated away from its origin (Dai et al., 2020), this is not the case at the ASP. 604 \nInstead, ice formed by the ASP often remains in the ASP study area for months (Video S1, Fig. 3). In fact, this po-\n605 \nlynya-produced ice does not consistently flow in a direction away from the polynya. While its overall direction is \n606 \nwestward, away from the polynya, the ice ‘heaves’ and temporarily flows ‘backward’ (‘back-fills’), as has also been \n607 \nobserved at the Mertz Glacier Polynya (Massom et al., 2017). The ice also gets ‘stuck’ in the region, particularly \n608 \naround grounded icebergs, which are grounded in areas where there are topographic highs in the sea floor (Fig. 2a, \n609 \ne). 610 \n \nThe tendency of ice produced by the ASP to remain in the vicinity of the polynya for prolonged periods, \n611 \nbecome stuck, and sometimes flow back eastward towards the site of its formation, influences the polynya and ice \n612 \nproduction in two key ways. 4.5 Broader SIC 637 \nBecause some ice has become stuck over the bathymetric high, when adjacent ice drifts away, a polynya opens. This \n638 \nfeature again highlights the significance of the bathymetry of the region for sea ice production and dynamics. 639 In line with previous studies of the ASP we find that the ASP is an important site of ice production \n640 \nthroughout the winter. Our estimates of annual ice production for 2019 and the mean for 2017-20 fall within the \n641 \nrange of predictions by Tamura et al. (2008; 2016) and Ohshima et al. (2017) for 1992-2001/2003-10/2013-15 (90 - \n642 \n117km3) (Fig. 6). Our estimates for total annual ice production for 2017 (139 km3) and 2018 (121 km3) are higher \n643 \nthan the highest estimate of those studies, while for 2020 (80 km3) it is lower. This suggests no significant trend in \n644 \ninternannual ice production can be discerned from comparing the period of our study to these previous studies. 645 \nSome caution must be used in this comparison, however, because those studies include March in ice production cal-\n646 \nculations and our ASP study area does not exactly correspond to theirs. 647 We find that the shoulder seasons of April/May and September/October are particularly important for ice \n648 \nproduction due to the higher polynya area at these times, accounting for 36% and 42% of the annual ice production, \n649 \nrespectively. This is particularly the case in 2017, when an exceptionally high open polynya area in the summer, due \n650 \nto low icepack conditions (discussed below), continues into the winter period (Figs. 4-6). However, we show that at \n651 \nleast some polynya area opens and some ice production occurs throughout the whole winter (Fig. 5-6). Additionally \n652 \nthere can be spikes in polynya area and ice production in the deepest winter months. Most notably, polynya area \n653 \nreaches 26 631 km2 in June. Such isolated, winter events are not reflected in the daily mean for the whole study pe-\n654 \nriod, but only when analyzing daily changes for each year (Fig. 5), highlighting the importance of analyzing polynya \n655 \narea and ice production at the daily scale to discern important polynya dynamics. 656 When comparing our results to those of Cheng et al. (2017), who used the same method for calculating ice \n657 \nproduction at the Ross Ice Shelf Polynya, we find that the ASP produces substantially less ice. 4.5 Broader SIC 619 We also note that the ASP forms along the coast westward off a chain of icebergs that extend from the \n620 \nThwaites Iceberg Tongue. While some polynyas, such as the Ross Ice Shelf polynya, form off and away from the \n621 \ncoast, the majority of Antarctic polynyas have been shown to form westward off glacier ice tongues or protruding \n622 \nfast ice (Nihashi et al., 2017). In the case of the ASP, its location and orientation is determined by the presence of \n623 \nthe ‘Iceberg Chain’, which is in turn determined by the presence of a bathymetric high (Figs. 1-2). Stammerjohn et \n624 \nal. (2015) refer to the polynya as forming off a ‘fast ice tongue’ but we prefer to refer to the ‘Iceberg Chain’ as the \n625 \neastern boundary. While a section of fast ice exists amongst, and adjacent to, a section of the southern part of the \n626 \nIceberg Chain, the extent of this fast ice varies and it only ever extends along a portion of the Iceberg Chain. The \n627 \nIceberg Chain remains virtually the same length throughout our observations. The polynya consistently forms off the \n628 25 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Iceberg Chain regardless of the state or extent of the fast ice. This means that, unlike polynyas that form off variable \n629 \nfast ice (Nihashi et al., 2017), the fundamental morphology of the ASP remains stable through the period and is \n630 \nlikely to remain so. We also note that at no point do we observe significant portions of icepack to ‘break through’ \n631 \nthe Iceberg Chain from east, regardless of the state of fast ice extent or conditions, and thus the icebergs and the \n632 \nbathymetric high persistently ‘shield’ the polynya from icepack inflow. 633 Another notable feature of the ASP is the development of a ‘secondary polynya’ during the winter (Fig. 634 \n2a), where ice production also takes place. This is a polynya that forms within the ASP study area, in an area that is \n635 \nusually part of main ASP during the summer, but it is not typically congruent with the ‘main’ polynya during the \n636 \nwinter. The polynya forms at the site of the ‘Central Grounded Icebergs’ and associated ‘stuck’, transient fast ice. 4.5 Broader SIC This westward flow of the ice away \n672 \nfrom the polynya is likely, in this section, primarily due to the prevalence of westward ocean currents in the region \n673 \n(St-Laurent et al., 2019), rather than wind which has a mean southerly direction (Fig. 7). These ocean currents have \n674 \nbeen shown to carry icebergs away from our study region, westward through the Amundsen Sea and into the Ross \n675 \nSea (Koo et al., in review). Broader prevailing easterly winds likely play a dominant role in sea ice produced by the \n676 \nASP eventually drifting to the Ross Sea, as part of a coastal band of westward ice drift (Assmann et al., 2005). The \n677 \nfact that the polynya-produced ice remains by the coast may also be influenced by the inflow of older, thicker ice-\n678 \npack into our study area. Icepack appears to flow into the region from the Bellingshausen Sea (Video S1, S2) and \n679 \nflows parallel to the ASP-produced ice, potentially playing some role in ‘trapping’ the ice by the coast. The west-\n680 \nward flow of the ice suggests that the level of ice production in the ASP is significant for the adjacent sector of the \n681 \nAmundsen Sea and the Ross Sea. 682 During the summer we observe the ASP to behave in a similar way in 2016/17 - 2020/21 as Stammerjohn \n683 \net al. (2015) showed for the period 1979 - 2014. As they did, we find the polynya to open every summer during our \n684 \nstudy period. We do not note any shift in the location of opening, with the location remaining in the same place as \n685 \nStammerjohn et al. (2015) noted that it had shifted to in 1992/93. As they did for the years 1992, 1993, 1995, 1997, \n686 \n2003, and 2010, we also note that in 2016/17, there is no icepack adjacent to the ASP in the north and west. This is \n687 \ndue to limited advection from the Bellingshausen Sea and Pine Island Polynya, and it causes the ASP to become \n688 \ncongruent with the open ocean (Fig. 11). This year was noted as a year of unprecedented springtime retreat and low \n689 \nsea ice concentration for Antarctic sea ice, and was associated with a series of record atmospheric circulation anom-\n690 \nalies and sea surface temperatures (Turner et al., 2017). 4.5 Broader SIC 670 \n \nWe also note the polynya-produced ice leaves our study area and enters the adjacent sector of the Amund-\n671 \nsen Sea to the west, rather than traveling away from the coast after formation. This westward flow of the ice away \n672 \nfrom the polynya is likely, in this section, primarily due to the prevalence of westward ocean currents in the region \n673 \n(St-Laurent et al., 2019), rather than wind which has a mean southerly direction (Fig. 7). These ocean currents have \n674 \nbeen shown to carry icebergs away from our study region, westward through the Amundsen Sea and into the Ross \n675 \nSea (Koo et al., in review). Broader prevailing easterly winds likely play a dominant role in sea ice produced by the \n676 \nASP eventually drifting to the Ross Sea, as part of a coastal band of westward ice drift (Assmann et al., 2005). The \n677 \nfact that the polynya-produced ice remains by the coast may also be influenced by the inflow of older, thicker ice-\n678 \npack into our study area. Icepack appears to flow into the region from the Bellingshausen Sea (Video S1, S2) and \n679 \nflows parallel to the ASP-produced ice, potentially playing some role in ‘trapping’ the ice by the coast. The west-\n680 \nward flow of the ice suggests that the level of ice production in the ASP is significant for the adjacent sector of the \n681 \nAmundsen Sea and the Ross Sea. 682 line, and is only typically limited in this spatial dimension by weather/oceanographic conditions. Another compara-\n665 \ntive limit on the ASP’s ice production is the previously discussed tendency for polynya-produced ice to inhibit fur-\n666 \nther opening of the polynya due to blockages and reversals in ice drift. This process could also partly explain why \n667 \nCheng et al. (2017) found ice production to remain relatively consistent throughout the winter for the Ross Ice Shelf \n668 \nPolynya, whereas we find ice production for the ASP in June-August to be much lower than in the shoulder months \n669 \nof April/May and September/October. 670 We also note the polynya-produced ice leaves our study area and enters the adjacent sector of the Amund-\n671 \nsen Sea to the west, rather than traveling away from the coast after formation. 4.5 Broader SIC Between 2003 and \n658 \n2015 they found ice production for the Ross Ice Shelf Polynya was between 164 and 313 km3 (also for April - Octo-\n659 \nber) compared to 80 to 139 km3 (this study). This is in line with other studies that compare the two polynyas \n660 \n(Tamura et al. 2008; 2016; Nihashi et al. 2017). We suggest that one limit on polynya area and ice production for the \n661 \nASP compared to the larger Ross Ice Shelf Polynya, is that the ASP typically forms off the Iceberg Chain. The Ice-\n662 \nberg Chain has a stable length of ~190 km, limited by the length of the seafloor sill on which it is grounded, and is \n663 \nan upper limit on the polynya in one dimension. The Ross Ice Shelf Polynya, on the other hand, forms off a coast-\n664 26 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. line, and is only typically limited in this spatial dimension by weather/oceanographic conditions. Another compara-\n665 \ntive limit on the ASP’s ice production is the previously discussed tendency for polynya-produced ice to inhibit fur-\n666 \nther opening of the polynya due to blockages and reversals in ice drift. This process could also partly explain why \n667 \nCheng et al. (2017) found ice production to remain relatively consistent throughout the winter for the Ross Ice Shelf \n668 \nPolynya, whereas we find ice production for the ASP in June-August to be much lower than in the shoulder months \n669 \nof April/May and September/October. 670 line, and is only typically limited in this spatial dimension by weather/oceanographic conditions. Another compara-\n665 \ntive limit on the ASP’s ice production is the previously discussed tendency for polynya-produced ice to inhibit fur-\n666 \nther opening of the polynya due to blockages and reversals in ice drift. This process could also partly explain why \n667 \nCheng et al. (2017) found ice production to remain relatively consistent throughout the winter for the Ross Ice Shelf \n668 \nPolynya, whereas we find ice production for the ASP in June-August to be much lower than in the shoulder months \n669 \nof April/May and September/October. 6. Conclusions \n707 Focusing on the summers of 2016/17 - 2020/21 and the winters of 2017 - 2020, we present the first detailed \n708 \nstudy of year-round variations in the Amundsen Sea Polynya’s behavior, area, and ice production. In particular, we \n709 \ntake advantage of the recent availability of Sentinel-1 SAR imagery to qualitatively assess the dynamics of the po-\n710 \nlynya through the whole year. 711 Our findings agree with previous studies of earlier periods in finding that the ASP produces a substantial \n712 \namount of ice through the winter, with some inter-annual variation. We add that the shoulder seasons of April/May \n713 \nand September/October dominate winter ice production, contributing a combined 78%. However, large polynya \n714 \nevents, often associated with high winds, can occur throughout the winter, promoting significant ice production. 715 The ASP opens each summer in November and closes in March or early April, with peak area typically oc-\n716 \ncurring in January. We find that broader regional sea ice conditions can play an important role in the polynya in \n717 \nsummer, with the record-low sea ice extent in 2016/17 causing the ASP to become part of the open ocean to the \n718 \nnorth and join with the Pine Island Polynya to the east. 719 Through our qualitative assessment we identify that the ASP behaves in a distinct manner. The polynya \n720 \ntypically forms in a westward direction off a persistent chain of grounded icebergs that are grounded along a bathy-\n721 \nmetric high. Ice produced by the polynya is not efficiently evacuated from the site as with other polynyas such as the \n722 \nRoss Ice Shelf Polynya. Instead it stays within the study site, typically for months through the winter, sometimes \n723 \nbecoming stuck. This behavior is related to local topographic sea-floor highs which cause icebergs to become \n724 \ngrounded and ice to become stuck. At times another smaller ‘secondary polynya’ forms within the study area adja-\n725 \ncent to grounded icebergs. Relatedly, ice produced by the polynya does not consistently move away from the ASP, \n726 \ninstead ‘heaving’ and sometimes drifting back towards it, contributing to its closure and limiting ice production. Un-\n727 \nlike some other polynyas, the polynya-produced ice also drifts westward into other sectors, instead of north, away \n728 \nfrom the coast. These behaviors should be accounted for when considering the ASP’s influence on the region’s sea \n729 \nice, biology and oceanography. 4.5 Broader SIC These broader sea ice conditions caused the polynya to be \n691 \nopen in approximately the whole ASP study area through most of the summer in 2016/17, from late November to \n692 \nMarch. During this time there is also little-to-no distinction between the ASP and the neighboring Pine Island Po-\n693 \nlynya, other than the presence of the Iceberg Chain (Fig. 11a, Video S1). The effect of these extraordinary sea ice \n694 \nconditions in 2016/17 on the polynya in summer, and early winter as mentioned above, may offer insight into how \n695 \nthe ASP will behave more commonly in future if climate change makes such conditions more likely. 696 We note that while Stammerjohn et al. (2015) found the largest polynya area to be February in all but two \n697 \nyears during 1979 - 2014, we find the polynya area to be highest in January in each year apart from 2016/17 (when it \n698 \nreaches the peak in November) (Fig. 4). Arrigo et al. (2012) also generally found the polynya area to increase until a \n699 \nlater peak in February for the years 1997/98 - 2009/10. While there should be some caution in directly comparing \n700 \nour results with those, due to varying datasets, methods and definitions of the study area, we suggest that a shift in \n701 27 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. the timing of maximum summer area would promote primary productivity in the polynya. Arrigo et al. (2012) found \n702 \nprimary productivity (per unit area) to typically peak in January, and to be declining by the time of the polynya area \n703 \npeak. If the polynya reaches a higher area at an earlier time, when primary productivity is higher, we suggest the \n704 \npotential for primary productivity may be larger during our study period. 705 \n706 the timing of maximum summer area would promote primary productivity in the polynya. Arrigo et al. (2012) found \n702 \nprimary productivity (per unit area) to typically peak in January, and to be declining by the time of the polynya area \n703 \npeak. If the polynya reaches a higher area at an earlier time, when primary productivity is higher, we suggest the \n704 \npotential for primary productivity may be larger during our study period. 705 \n \n706 \n6. 4.5 Broader SIC Conclusions \n707 \n \nFocusing on the summers of 2016/17 - 2020/21 and the winters of 2017 - 2020, we present the first detailed \n708 \nstudy of year-round variations in the Amundsen Sea Polynya’s behavior, area, and ice production. In particular, we \n709 \ntake advantage of the recent availability of Sentinel-1 SAR imagery to qualitatively assess the dynamics of the po-\n710 \nlynya through the whole year. 711 \n \nOur findings agree with previous studies of earlier periods in finding that the ASP produces a substantial \n712 \namount of ice through the winter, with some inter-annual variation. We add that the shoulder seasons of April/May \n713 \nand September/October dominate winter ice production, contributing a combined 78%. However, large polynya \n714 \nevents, often associated with high winds, can occur throughout the winter, promoting significant ice production. 715 \n \nThe ASP opens each summer in November and closes in March or early April, with peak area typically oc-\n716 \ncurring in January. We find that broader regional sea ice conditions can play an important role in the polynya in \n717 \nsummer, with the record-low sea ice extent in 2016/17 causing the ASP to become part of the open ocean to the \n718 \nnorth and join with the Pine Island Polynya to the east. 719 \n \nThrough our qualitative assessment we identify that the ASP behaves in a distinct manner. The polynya \n720 \ntypically forms in a westward direction off a persistent chain of grounded icebergs that are grounded along a bathy-\n721 \nmetric high. Ice produced by the polynya is not efficiently evacuated from the site as with other polynyas such as the \n722 \nRoss Ice Shelf Polynya. Instead it stays within the study site, typically for months through the winter, sometimes \n723 \nbecoming stuck. This behavior is related to local topographic sea-floor highs which cause icebergs to become \n724 \ngrounded and ice to become stuck. At times another smaller ‘secondary polynya’ forms within the study area adja-\n725 \ncent to grounded icebergs. Relatedly, ice produced by the polynya does not consistently move away from the ASP, \n726 \ninstead ‘heaving’ and sometimes drifting back towards it, contributing to its closure and limiting ice production. Un-\n727 \nlike some other polynyas, the polynya-produced ice also drifts westward into other sectors, instead of north, away \n728 \nfrom the coast. These behaviors should be accounted for when considering the ASP’s influence on the region’s sea \n729 \nice, biology and oceanography. 4.5 Broader SIC 730 \n \nGiven temporal and spatial gaps in Sentinel-1 SAR’s coverage, we do not find that it can replace passive \n731 \nmicrowave or sea ice concentration datasets for analyzing daily changes in polynya area or ice production. However, \n732 \nwe find that the ability to directly observe and qualitatively analyze the polynya at a high spatial and temporal reso-\n733 \nlution, year-round, with Sentinel-1 imagery provides important insights that are not possible with those other da-\n734 \ntasets. We also note that it is sometimes possible to visually-track ice created by particular polynya events for sev-\n735 \neral months as it drifts. Employing ice-tracking algorithms to track ice produced by polynya events in Sentinel-1 \n736 \nimages, with measurements of ice thickness (e.g. from satellite altimetry), could help further quantify ice production \n737 \nby polynyas and extract more potential from Sentinel-1 datasets. 738 6. Conclusions \n707 We thank the University of Bremen \n767 \nSea Ice Remote Sensing Group for processing and making available their sea ice concentration product and the Japa-\n768 \nnese Space Exploration Agency for launching and managing AMSR2. We also acknowledge the ECMWF, NSIDC \n769 \nand NASA for freely-available data. Among others, the free, open-source software packages QGIS, GDAL, Python, \n770 \nNumPy and MatPlotLib were invaluable in this research. We also thank the GIS and StackOverflow StackExchange \n771 \ncommunities for resources that aided data processing. 772 \n \n773 \nReferences \n774 \nArmstrong, T.: World meteorological organization: WMO sea-ice nomenclature. Terminology, codes and illustrated \n775 \nglossary. J. Glaciol., 11, 148–149. https://doi.org/10.3189/S0022143000022577, 1972. 776 Code and Data Availability \n739 \nCode for data processing and production of figures and videos is available at \n740 \nhttps://github.com/geogeordie/AmundsenSeaPolynyaPaper. All processing was done with freely available software, \n741 \nand all data is freely available. Sentinel-1 images were processed in Google Earth Engine or downloaded from: \n742 \nasf.alaska.edu. BedMachine Antarctica V2 was downloaded from: https://nsidc.org/data/nsidc-0756. Sea ice concen-\n743 \ntration data was downloaded from: http://seaice.uni-bremen.de/. ERA5 climate data was downloaded from: \n744 \nhttps://cds.climate.copernicus.eu. The MODIS image used for Fig. 1b was downloaded from: \n745 \nhttps://worldview.earthdata.nasa.gov/ \n746 \n \n747 \nVideo Supplement \n748 \nVideo S1: https://doi.org/10.5281/zenodo.5179444 \n749 \nVideo S2: https://doi.org/10.5281/zenodo.5179509 \n750 \nVideo S3: https://doi.org/10.5281/zenodo.5179590 \n751 \n \n752 \nAuthor Contributions \n753 \nGJM primarily conceived the study, processed and analyzed all data and produced all figures. SFA and AMM-N \n754 \nalso contributed to the design of the study and all authors discussed the results and were involved in editing the man-\n755 \nuscript. 756 \n \n757 \nCompeting Interests \n758 \nThe authors declare that they have no conflict of interest. 759 \n \n760 \nFinancial Support \n761 \nThis work was supported by NASA grant #80NSSC19M0194. 762 \n \n763 \nAcknowledgements \n764 \nWe gratefully acknowledge the European Space Agency for making available Sentinel-1 data and the SNAP \n765 \ntoolbox, and Google Earth Engine for hosting and processing Sentinel-1 data. We acknowledge the free package \n766 \nQuantarctica, developed by the Norwegian Polar Institute, for use in Fig. 1a. We thank the University of Bremen \n767 \nSea Ice Remote Sensing Group for processing and making available their sea ice concentration product and the Japa-\n768 \nnese Space Exploration Agency for launching and managing AMSR2. We also acknowledge the ECMWF, NSIDC \n769 \nand NASA for freely-available data. Among others, the free, open-source software packages QGIS, GDAL, Python, \n770 \nNumPy and MatPlotLib were invaluable in this research. 6. Conclusions \n707 730 Given temporal and spatial gaps in Sentinel-1 SAR’s coverage, we do not find that it can replace passive \n731 \nmicrowave or sea ice concentration datasets for analyzing daily changes in polynya area or ice production. However, \n732 \nwe find that the ability to directly observe and qualitatively analyze the polynya at a high spatial and temporal reso-\n733 \nlution, year-round, with Sentinel-1 imagery provides important insights that are not possible with those other da-\n734 \ntasets. We also note that it is sometimes possible to visually-track ice created by particular polynya events for sev-\n735 \neral months as it drifts. Employing ice-tracking algorithms to track ice produced by polynya events in Sentinel-1 \n736 \nimages, with measurements of ice thickness (e.g. from satellite altimetry), could help further quantify ice production \n737 \nby polynyas and extract more potential from Sentinel-1 datasets. 738 28 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Code and Data Availability \n739 \nCode for data processing and production of figures and videos is available at \n740 \nhttps://github.com/geogeordie/AmundsenSeaPolynyaPaper. All processing was done with freely available software, \n741 \nand all data is freely available. Sentinel-1 images were processed in Google Earth Engine or downloaded from: \n742 \nasf.alaska.edu. BedMachine Antarctica V2 was downloaded from: https://nsidc.org/data/nsidc-0756. Sea ice concen-\n743 \ntration data was downloaded from: http://seaice.uni-bremen.de/. ERA5 climate data was downloaded from: \n744 \nhttps://cds.climate.copernicus.eu. The MODIS image used for Fig. 1b was downloaded from: \n745 \nhttps://worldview.earthdata.nasa.gov/ \n746 \n \n747 \nVideo Supplement \n748 \nVideo S1: https://doi.org/10.5281/zenodo.5179444 \n749 \nVideo S2: https://doi.org/10.5281/zenodo.5179509 \n750 \nVideo S3: https://doi.org/10.5281/zenodo.5179590 \n751 \n \n752 \nAuthor Contributions \n753 \nGJM primarily conceived the study, processed and analyzed all data and produced all figures. SFA and AMM-N \n754 \nalso contributed to the design of the study and all authors discussed the results and were involved in editing the man-\n755 \nuscript. 756 \n \n757 \nCompeting Interests \n758 \nThe authors declare that they have no conflict of interest. 759 \n \n760 \nFinancial Support \n761 \nThis work was supported by NASA grant #80NSSC19M0194. 762 \n \n763 \nAcknowledgements \n764 \nWe gratefully acknowledge the European Space Agency for making available Sentinel-1 data and the SNAP \n765 \ntoolbox, and Google Earth Engine for hosting and processing Sentinel-1 data. We acknowledge the free package \n766 \nQuantarctica, developed by the Norwegian Polar Institute, for use in Fig. 1a. 6. Conclusions \n707 We also thank the GIS and StackOverflow StackExchange \n771 \ncommunities for resources that aided data processing. 772 \n \n773 \nReferences \n774 \nArmstrong, T.: World meteorological organization: WMO sea-ice nomenclature. Terminology, codes and illustrated \n775 \nglossary. J. Glaciol., 11, 148–149. https://doi.org/10.3189/S0022143000022577, 1972. 776 Armstrong, T.: World meteorological organization: WMO sea-ice nomenclature. Terminology, codes and illustrated \n775 \nglossary. J. Glaciol., 11, 148–149. https://doi.org/10.3189/S0022143000022577, 1972. \n776 References \n774 Armstrong, T.: World meteorological organization: WMO sea-ice nomenclature. Terminology, codes and illustrated \n775 \nglossary. J. Glaciol., 11, 148–149. https://doi.org/10.3189/S0022143000022577, 1972. 776 29 777 \nArrigo, K. R., van Dijken, G. L., and Bushinsky, S.: Primary production in the Southern Ocean, 1997–2006, J. Ge-\n778 \nophys. Res., 113, C08004. https://doi.org/10.1029/2007JC004551, 2008a. 779 \n780 780 \nArrigo, K. R., van Dijken, G. L., and Long, M.: Coastal Southern Ocean: A strong anthropogenic CO2 sink. Ge-\n781 \nophys. Res. Lett., 35, L21602. https://doi.org/10.1029/2008GL035624, 2008b. 782 \n783 783 \nArrigo, K. R., Lowry, K. E., and van Dijken, G. L.:Annual changes in sea ice and phytoplankton in polynyas of the \n784 \nAmundsen Sea, Antarctica. Deep Sea Research, II 71–76: 5–15. https://doi.org/10.1016/j.dsr2.2012.03.006, 2012. 785 \n786 786 \nArrigo, K. R. and van Dijken, G. L.: Phytoplankton dynamics within 37 Antarctic coastal polynya systems. J. Ge-\n787 \nophys. Res., 108, 3271 https://doi.org/10.1029/2002jc001739, 2003. 788 \n789 789 \nArthur, J. F, Stokes, C. R., Jamieson, S. S. R., Miles, B. W. J., Carr, J. R, and Leeson, A. A.: The triggers of the dis-\n790 \naggregation of Voyeykov Ice Shelf (2007), Wilkes Land, East Antarctica, and its subsequent evolution. J. Glaciol \n791 \nhttps://doi.org/10.1017/jog.2021.45, 2021. 792 \n793 793 \nAssmann, K. M., Hellmer, H. H., and Jacobs, S. S.: Amundsen Sea ice production and transport. J. Geophys. Res., \n794 \n110, C12013. https://doi.org/10.1029/2004JC002797, 2005. 795 \n796 796 \nBanwell, A. F., Willis, I. C., Macdonald, G. J., Goodsell, B., Mayer, D. P., Powell, A., and MacAyeal, D. R.: Calv-\n797 \ning and rifting on the McMurdo Ice Shelf, Antarctica. Ann. Glaciol., 58, 78–87. https://doi.org/10.1017/aog.2017.12, \n798 \n2017. 799 \n800 800 \nBett, D. T., Holland, P. R., Naveira Garabato, A. C., Jenkins, A., Dutrieux, P., Kimura, S., and Fleming, A.: The im-\n801 \npact of the Amundsen Sea freshwater balance on ocean melting of the West Antarctic Ice Sheet. J. Geophys. Res.: \n802 \nOceans, 125, e2020JC016305. https://doi.org/10.1029/2020JC016305, 2020. 803 \n804 804 \nBracegirdle, T. J.: Climatology and recent increase of westerly winds over the Amundsen Sea derived from six rea-\n805 \nnalyses. Int. J. Climatol., 33, 843–851. https://doi.org/10.1002/joc.3473, 2013. 806 \n807 807 \nBromwich, D. H. and Kurtz, D. D.: Katabatic wind forcing of the Terra Nova Bay polynya. J. Geophys. Res., 89, \n808 \n3561. https://doi.org/10.1029/JC089iC03p03561, 1984. 809 \n810 810 \nBromwich, D., Liu, Z., Rogers, A. N., and Van Woert, M. L.: Winter atmospheric forcing of the Ross Sea polynya. 811 \nOcean ICE Atmos. Int. Antarct. Cont. Margin, 75, 101-133. References \n774 https://doi.org/10.1029/AR075p0101, 1998. 812 \n813 813 \nBromwich, D. H., Carrasco, J. F., Liu, Z., and Tzeng, R. Y.: Hemispheric atmospheric variations and oceanographic \n814 \nimpacts associated with katabatic surges across the ross ice shelf, Antarctica. J. Geophys. Res.-Atmos, 98, 13045-\n815 \n13062. https://doi.org/10.1029/93JD00562, 1993. 816 \n817 817 \nCheng, Z., Pang, X., Zhao, X., and Tan, C.: Spatio-Temporal Variability and Model Parameter Sensitivity Analysis \n818 \nof Ice Production in Ross Ice Shelf Polynya from 2003 to 2015. Remote Sensing, 9, 934. 819 \nhttps://doi.org/10.3390/rs9090934, 2017. 820 \n821 821 \nCheng, Z., Pang, X., Zhao, X., and Stein, A.: Heat Flux Sources Analysis to the Ross Ice Shelf Polynya Ice Produc-\n822 \ntion Time Series and the Impact of Wind Forcing, Remote Sensing, 11, 188. https://doi.org/10.3390/rs11020188, \n823 \n2019. 824 \n825 825 \nDai, L., Xie, H., Ackley, S. F., and Mestas-Nuñez, A. M.: Ice Production in Ross Ice Shelf Polynyas during 2017–\n826 \n2018 from Sentinel–1 SAR Images. Remote Sensing, 12, 1484. https://doi.org/10.3390/rs12091484, 2020. 827 \n828 828 \nDrucker, R., Martin, S., and Kwok, R..: Sea ice production and export from coastal polynyas in the Weddell and \n829 \nRoss Seas. Geophys. Res. Lett., 38, L1705. https://doi.org/10.1029/2011GL048668, 2011. 830 \n831 Drucker, R., Martin, S., and Kwok, R..: Sea ice production and export from coastal polynyas in the Weddell and \n829 \nRoss Seas. Geophys. Res. Lett., 38, L1705. https://doi.org/10.1029/2011GL048668, 2011. 830 \n831 831 \nGreene, C. E., Young, D. A., Gwyther, D. E., Galton-Fenzi, B. E., and Blankenship, D. D.: Seasonal dynamics of \n832 831 \nGreene, C. E., Young, D. A., Gwyther, D. E., Galton-Fenzi, B. E., and Blankenship, D. D.: Seasonal dynamics of \n832 30 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Totten Ice Shelf controlled by sea ice buttressing. Cryosphere, 12, 2869-2882. https://doi.org/10.5194/tc-12-2869-\n833 \n2018, 2018. 834 \n835 835\nGrossmann, S. and Dieckmann, G. S.: Bacterial Standing Stock, Activity, and Carbon Production during Formation \n836 \nand Growth of Sea–Ice in the Weddell Sea, Antarctica. Appl. Environ. Microb., 60, 2746–2753. 837 \nhttps://doi.org/10.1128/aem.60.8.2746-2753.1994, 1994. 838 \n839 Hersbach, H., Bell, B., Berrisford, P., Biavati, G., Horányi, A., Muñoz Sabater, J., Nicolas, J., Peubey, C., Radu, R., \n840 \nRozum, I., Schepers, D., Simmons, A., Soci, C., Dee, D., and Thépaut, J.-N.: ERA5 monthly averaged data on single \n841 \nlevels from 1979 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS). References \n774 (Accessed on < \n842 \n01-04-2021), https://doi.org/10.24381/cds.f17050d7, 2019. 843 \n844 844 \nIMBIE team: Mass balance of the Antarctic ice sheet from 1992 to 2017. Nature 558, 219-222. 845 \nhttps://doi.org/10.1038/s41586-018-0179-y, 2018. 846 \n847 847 \nIto, M., Ohshima, K. I., Fukamachi, Y., Mizuta, G., Kusumoto, Y., and Nishioka, J.: Observations of frazil ice for-\n848 \nmation and upward sediment transport in the Sea of Okhotsk: A possible mechanism of iron supply to sea ice. J. Ge-\n849 \nophys. Res.–Oceans, 122, 788–802. https://doi.org/10.1002/2016JC012198, 2017. 850 \n851 851 \nKimura, N. and Wakatsuchi, M.: Increase and decrease of sea ice area in the Sea of Okhotsk: Ice production in \n852 \ncoastal polynyas and dynamic thickening in convergence zones. J. Geophys. Res.-Oceans, 109, C09S03. 853 \nhttps://doi.org/10.1029/2003JC001901, 2004. 854 \n855 855 \nKitade, Y., Shimada, K., Tamura, T., Williams, G. D., Aoki, S., Fukamachi, Y., Roquet, F., Hindell, M., Ushio, S., \n856 \nand Ohshima, K. I.: Antarctic Bottom Water production from the Vincennes Bay Polynya, East Antarctica. Ge-\n857 \nophys. Res. Lett., 41, 3528–3534. https://doi.org/10.1002/2014GL059971, 2014. 858 \n859 859 \nKoo, Y., Xie, H., Ackley, S. F., Mestas-Nuñez, A. M., Macdonald, G. J., and Hyun, C.-U.: Semi-automated tracking \n860 \nof iceberg B43 using Sentinel-1 SAR images via Google Earth Engine. Cryosphere Discuss., \n861 \nhttps://doi.org/10.5194/tc-2021-131, 2021 [in review] \n862 \n863 Krumpen, T., Hölemann, J. A., Willmes, S., Morales Maqueda, M. A., Busche, T., Dmitrenko, I. A., Gerdes, R., \n864 \nHaas, C., Heinemann, G., Hendricks, S., Kassens, H., Rabenstein, L., and Schröder, D.: Sea ice production and wa-\n865 \nter mass modification in the eastern Laptev Sea. J. Geophys. Res.-Oceans, 116, C05014. 866 \nhttps://doi.org/10.1029/2010JC006545, 2011. 867 \n868 868 \nMarzocchi, A. and Jansen, M. F..: Global cooling linked to increased glacial carbon storage via changes in Antarctic \n869 \nsea ice, Nat. Geosci., 12, 1001-1005. https://doi.org/10.1038/s41561-019-0466-8, 2019. 870 \n871 871 \nMassom, R. A., Hill, K. L., Lytle, V. I., Worby, A.P., Paget, M. J., and Allison, I.: Effects of regional fast-ice and \n872 \niceberg distributions on the behaviour of the Mertz Glacier polynya, East Antarctica. Ann. Glaciol., 33, 391-398. 873 \nhttps://doi.org/10.3189/172756401781818518, 2001. 874 \n875 875 \nMassom, R. A., Scambos, T. A., Bennetts, L. G., Reid, P., Squire, V. A., and Stammerjohn, S. E.: Antarctic ice shelf \n876 \ndisintegration triggered by sea ice loss and ocean swell. Nature, 558, 383-389. https://doi.org/10.1038/s41586-018-\n877 \n0212-1, 2018. 885 \nMorales Maqueda, M. A., Willmott, A. J., and Biggs, N. R. T.: Polynya dynamics: a review of observations and \n886 \nmodeling, Review of Geophysics, 42, RG1004. https://doi.org/10.1029/2002RG000116, 2004. \n887 \n888 References \n774 878 \n879 Matsuoka, K., Skoglund, A., Roth, G., de Pomereu, J., Griffiths, H., Headland, R., Herried, B., Katsumata, K., Le \n880 \nBrocq, A., Licht, K., Morgan, F., Neff, P. D., Ritz, C., Scheinert, M., Tamura, T., Van de Putte, A., van den Broeke, \n881 \nM., von Deschwanden, A., Deschamps-Berger, C., Van Liefferinge, B., Tronstad, S., and Melvær, Y.: Quantarctica, \n882 \nan integrated mapping environment for Antarctica, the Southern Ocean, and sub-Antarctic islands. Environ. Modell. 883 \nSoftw., 140, 105015. https://doi.org/10.1016/j.envsoft.2021.105015, 2021. 884 \n885 885 \nMorales Maqueda, M. A., Willmott, A. J., and Biggs, N. R. T.: Polynya dynamics: a review of observations and \n886 \nmodeling, Review of Geophysics, 42, RG1004. https://doi.org/10.1029/2002RG000116, 2004. 887 Morales Maqueda, M. A., Willmott, A. J., and Biggs, N. R. T.: Polynya dynamics: a review of observations and \n886 31 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. CC BY 4.0 License. Morelli, S. and Parmiggiani, F.: Wind over Terra Nova Bay (Antarctica) during a polynya event: Eta model simula-\n889 \ntions and satellite microwave observations. Eur. Phys. J. Plus 128, 135 (2013). https://doi.org/10.1140/epjp/i2013-\n890 \n13135-8, 2013. 891 \n892 892 \nMorlighem, M.: MEaSUREs BedMachine Antarctica, Version 2. Boulder, Colorado USA. NASA National Snow \n893 \nand Ice Data Center Distributed Active Archive Center. https://doi.org/10.5067/C2GFER6PTOS4, 2019. [Date Ac-\n894 \ncessed: 01-04-2021]. 895 \n896 Nihashi, S., and Ohshima, K. I.: Circumpolar Mapping of Antarctic Coastal Polynyas and Landfast Sea Ice: Rela-\n897 \ntionship and Variability, J. Climate, 28, 3650-3670. https://doi.org/10.1175/JCLI-D-14-00369.1, 2015. 898 \n899 899 \nNihashi, S., Ohshima, K. I., and Tamura, T.: Sea-Ice Production in Antarctic Coastal Polynyas Estimated From \n900 \nAMSR2 Data and Its Validation Using AMSR–E and SSM/I–SSMIS Data. IEEE JSTARS, 10, 3912-3922. 901 \nhttps://doi.org/10.1109/JSTARS.2017.2731995, 2017. 902 \n903 903 \nOrsi, A. H., Johnson, G. C., and Bullister, J. L/.: Circulation, mixing, and production of Antarctic Bottom Water. 904 \nProg. Oceanogr., 43, 55-109. https://doi.org/10.1016/S0079-6611(99)00004-X, 2019. 905 \n906 Ohshima, K. I., Fukamachi, Y., Williams, G. D., Nihashi, S., Roquet, F., Kitade, Y., Tamura, T., Hirano, D., Her-\n907 \nraiz-Borreguero, L., Field, I., Hindell, M., Aoki, S. and Wakatsuchi, M. L.: Antarctic Bottom Water production by \n908 \nintense sea–ice formation in the Cape Darnley polynya. Nat. Geosci., 6, 235-240. https://doi.org/10.1038/ngeo1738, \n909 \n2013. References \n774 CC BY 4.0 License. the period of 1979–2014. Elementa: Science of the Anthropocene, 3, 000055. https://doi.org/10.12952/journal. 944 \nmenta.000055, 2015. 945 \n \n946 \nSt-Laurent, P., Yager, P. L., Sherrell, R. M., Stammerjohn, S. E., and Dinniman, M. S.: Pathways and supply o\n947 \nsolved iron in the Amundsen Sea (Antarctica), J. Geophys. Res.-Oceans, 122, 7135–7162, https://doi.org/10.10\n948 \n2017JC013162, 2017. 949 \n \n950 \nSt-Laurent, P, Yager, P. L., Sherrell, R. M., Oliver, H., Dinniman, M. S., and Stammerjohn, S. E.: Modeling th\n951 \nsonal cycle of iron and carbon fluxes in the Amundsen Sea Polynya, Antarctica. J. Geophys. Res.: Oceans, 124\n952 \n1544–1565. https://doi.org/10.1029/2018JC014773, 2019. 953 \n \n954 \nSweeney, C.: The annual cycle of surface water CO2 and O2 in the Ross Sea: A model for gas exchange on the\n955 \ntinental shelves of Antarctica, in Biogeochemistry of the Ross Sea. Antarctica Research Series, 78, edited by D\n956 \nbar, R. B., and DiTullio, G. R., pp. 295-312, AGU, Washington, D. C., 2003. 957 \n \n958 \nTamura, T., Ohshima, K. I., and Nihashi, S.: Mapping of sea ice production for Antarctic coastal polynyas, Ge-\n959 \nophys. Res. Lett., 35, L07606. https://doi.org/10.1029/2007GL03290, 2008. 960 \n \n961 \nTamura, T., Williams, G. D., Fraser, A. D., and Ohshima, K. I.: Potential regime shift in decreased sea ice prod\n962 \ntion after the Mertz Glacier calving. Nat. Commun., 3, 826. https://doi.org/10.1038/ncomms1820, 2008. 963 \n \n964 \nTian, L., Xie, H., Ackley, S. F., Tang, J., Mestas-Nuñez, A. M., and Wang, X.: Sea-ice freeboard and thickness\n965 \nthe Ross Sea from airborne (IceBridge 2013) and satellite (ICESat 2003-2008) observations. Ann. Glaciol., 61(\n966 \n24-39. https://doi.org/10.1017/aog.2019.49, 2020. 967 \n \n968 \nWebber, B. G. M., Heywood, K. J., Stevens, D. P., Dutrieux, P., Povl Abrahamsen, E., Jenkins, A., Jacobs, S. S\n969 \nHa, H. K., Lee, S. H., and Kim, T. W.: Mechanisms driving variability in the ocean forcing of Pine Island Glac\n970 \nNat. Commun., 8, 14507. https://doi.org/10.1038/ncomms14507, 2017. 971 the period of 1979–2014. Elementa: Science of the Anthropocene, 3, 000055. https://doi.org/10.12952/journal.ele-\n944 \nmenta.000055, 2015. 945 \n946 946 \nSt-Laurent, P., Yager, P. L., Sherrell, R. M., Stammerjohn, S. E., and Dinniman, M. S.: Pathways and supply of dis-\n947 \nsolved iron in the Amundsen Sea (Antarctica), J. Geophys. Res.-Oceans, 122, 7135–7162, https://doi.org/10.1002/ \n948 \n2017JC013162, 2017. 949 \n9 0 St-Laurent, P, Yager, P. L., Sherrell, R. M., Oliver, H., Dinniman, M. S., and Stammerjohn, S. References \n774 910 \n911 911 \nPark, J., Kim, H.-C., Jo, Y.-H., Kidwell, A., and Hwang, J.: Multi-temporal variation of the ross sea polynya in re-\n912 \nsponse to climate forcings. Polar Res., 37, 1444891. https://doi.org/10.1080/17518369.2018.1444891, 2018. 913 \n914 914 \nParkinson, C. L.: A 40-y record reveals gradual Antarctic sea ice increases followed by decreases at rates far exceed-\n915 \ning the rates seen in the Arctic. P. Natl. Acad. Sci. U.S.A., 116, 14414-14423. 916 \nhttps://doi.org/10.1073/pnas.1906556116, 2019. 917 \n918 918 \nParmiggiani, F.: Fluctuations of Terra Nova Bay polynya as observed by active (ASAR) and passive (AMSR‐E) mi-\n919 \ncrowave radiometers, Int. J. Remote. Sens., 27:12, 2459 2467. https://doi.org/10.1080/01431160600554355, 2007. 920 \n921 921 \nPreußer, A., Heinemann, G., Willmes, S., and Paul, S.: Multi-Decadal Variability of Polynya Characteristics and Ice \n922 \nProduction in the North Water Polynya by Means of Passive Microwave and Thermal Infrared Satellite Imagery. 923 \nRemote Sensing, 7, 15844-15867. https://doi.org/10.3390/rs71215807, 2015. 924 \n925 925 \nQGIS.org: QGIS Geographic Information System. QGIS Association. http://www.qgis.org. 2021. 926 \n927 927 \nRack, W., Price, D., Haas, C., Langhorne, P. J., and Leonard, G. H.: Sea Ice Thickness in the Western Ross Sea. Ge-\n928 \nophys. Res. Lett., 48, e2020GL090866. https://doi.org/10.1029/2020GL090866, 2020. 929 \n930 930 \nRignot, E., Mouginot, J., Scheuchl, B., van den Broeke, M., van Wessem, M. J., and Morlighem, M.: Four decades \n931 \nof Antarctic ice sheet mass balance from 1979–2017. P. Natl. Acad. Sci. U.S.A. 116, 1095-1103. 932 \nhttps://doi.org/10.1073/pnas.1812883116, 2019. 933 \n934 934 \nSansiviero, M., Morales Maqueda, M. Á., Fusco, G., Aulicino, G., Flocco, D., and Budillon, G.: Modelling sea ice \n935 \nformation in the Terra Nova Bay polynya. J. Marine Syst., 166, 4-25. https://doi.org/10.1016/j.jmarsys.2016.06.013, \n936 \n2017. 937 \n938 938 \nSpreen, G. Kaleschke, G. L, and Heygster, G.: Sea ice remote sensing using AMSR-E 89 GHz channels. J. Geophys. 939 \nRes., 113, C02S03. https://doi.org/10.1029/2005JC003384, 2008. 940 \n941 941 \nStammerjohn, S. E., Maksym, T., Massom, R. A., Lowry, K. E., Arrigo, K. R., Yuan, X., Raphael, M., Randall-\n942 \nGoodwin, E. R., Sherrell, R. M., and Yager, P. L.: Seasonal sea ice changes in the Amundsen Sea, Antarctica, over \n943 Stammerjohn, S. E., Maksym, T., Massom, R. A., Lowry, K. E., Arrigo, K. R., Yuan, X., Raphael, M., Randall-\n942 \nG\nd i\nE R Sh\nll R M\nd Y\nP L S\nl\ni\nh\ni th A\nd\nS\nA t\nti\n943 32 https://doi.org/10.5194/tc-2021-250\nPreprint. Discussion started: 21 September 2021\nc⃝Author(s) 2021. References \n774 E.: Modeling the sea-\n951 \nsonal cycle of iron and carbon fluxes in the Amundsen Sea Polynya, Antarctica. J. Geophys. Res.: Oceans, 124, \n952 \n1544–1565. https://doi.org/10.1029/2018JC014773, 2019. 953 \n954 954 \nSweeney, C.: The annual cycle of surface water CO2 and O2 in the Ross Sea: A model for gas exchange on the con-\n955 \ntinental shelves of Antarctica, in Biogeochemistry of the Ross Sea. Antarctica Research Series, 78, edited by Dun-\n956 \nbar, R. B., and DiTullio, G. R., pp. 295-312, AGU, Washington, D. C., 2003. 957 \n958 958 \nTamura, T., Ohshima, K. I., and Nihashi, S.: Mapping of sea ice production for Antarctic coastal polynyas, Ge-\n959 \nophys. Res. Lett., 35, L07606. https://doi.org/10.1029/2007GL03290, 2008. 960 \n961 961 \nTamura, T., Williams, G. D., Fraser, A. D., and Ohshima, K. I.: Potential regime shift in decreased sea ice produc-\n962 \ntion after the Mertz Glacier calving. Nat. Commun., 3, 826. https://doi.org/10.1038/ncomms1820, 2008. 963 \n964 964 \nTian, L., Xie, H., Ackley, S. F., Tang, J., Mestas-Nuñez, A. M., and Wang, X.: Sea-ice freeboard and thickness in \n965 \nthe Ross Sea from airborne (IceBridge 2013) and satellite (ICESat 2003-2008) observations. Ann. Glaciol., 61(82), \n966 \n24-39. https://doi.org/10.1017/aog.2019.49, 2020. 967 \n968 968 \nWebber, B. G. M., Heywood, K. J., Stevens, D. P., Dutrieux, P., Povl Abrahamsen, E., Jenkins, A., Jacobs, S. S., \n969 \nHa, H. K., Lee, S. H., and Kim, T. W.: Mechanisms driving variability in the ocean forcing of Pine Island Glacier, \n970 \nNat. Commun., 8, 14507. https://doi.org/10.1038/ncomms14507, 2017. 971 Webber, B. G. M., Heywood, K. J., Stevens, D. P., Dutrieux, P., Povl Abrahamsen, E., Jenkins, A., Jacobs, S. S., \n969 \nHa, H. K., Lee, S. H., and Kim, T. W.: Mechanisms driving variability in the ocean forcing of Pine Island Glacier, \n970 \nNat. Commun., 8, 14507. https://doi.org/10.1038/ncomms14507, 2017. 971 33"