aaaTman commited on
Commit
de7dbf8
·
verified ·
1 Parent(s): 27ed971

Rename current model to v2.0.0.keras; add surface geopotential input and model card

Browse files
.gitattributes CHANGED
@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  uv_q_theta_e.keras filter=lfs diff=lfs merge=lfs -text
37
  nbs_square_dilation_trof_tt_no_inst.keras filter=lfs diff=lfs merge=lfs -text
 
 
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  uv_q_theta_e.keras filter=lfs diff=lfs merge=lfs -text
37
  nbs_square_dilation_trof_tt_no_inst.keras filter=lfs diff=lfs merge=lfs -text
38
+ v2.0.0.keras filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -2,9 +2,70 @@
2
  license: cc0-1.0
3
  language:
4
  - en
 
5
  tags:
6
  - meteorology
7
  - atmospheric
8
  - science
9
  - fronts
10
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
  license: cc0-1.0
3
  language:
4
  - en
5
+ library_name: keras
6
  tags:
7
  - meteorology
8
  - atmospheric
9
  - science
10
  - fronts
11
+ ---
12
+
13
+ # FrontFinder
14
+
15
+ Keras models that detect synoptic and mesoscale weather features (fronts, troughs, drylines) on a global 0.25° grid from numerical weather prediction fields.
16
+
17
+ ## Files
18
+
19
+ | File | Description |
20
+ |---|---|
21
+ | `v2.0.0.keras` | Current model (served at fronts.espr.ai). Previously named `nbs_square_dilation_trof_tt_no_inst.keras`. |
22
+ | `uv_q_theta_e.keras` | Earlier model, kept for reference. |
23
+ | `geopotential_at_surface_era5_0p25deg.npy` | Static surface geopotential (terrain) input required by `v2.0.0.keras`. |
24
+
25
+ ## v2.0.0
26
+
27
+ A 2D UNet3+ that takes a flat 28-channel input and outputs per-pixel class probabilities.
28
+
29
+ ### Inputs
30
+
31
+ Fields on the ECMWF IFS open-data 0.25° grid (721 x 1440). The model contains its own normalization (Rescaling) layer, so inputs are given in physical units.
32
+
33
+ Pressure-level variables, each at 1000, 925, 850, 700, 500 and 300 hPa (6 levels x 4 variables = 24 channels):
34
+ - equivalent potential temperature (theta-e), derived from IFS fields
35
+ - u component of wind
36
+ - v component of wind
37
+ - specific humidity
38
+
39
+ Single-level variables (4 channels):
40
+ - 10 m u wind
41
+ - 10 m v wind
42
+ - 2 m dewpoint temperature
43
+ - surface geopotential (m² s⁻², ERA5 convention, not height in meters)
44
+
45
+ Surface geopotential is not part of the IFS open-data real-time stream. It is supplied by `geopotential_at_surface_era5_0p25deg.npy`: float32, shape (721, 1440), latitude 90 to -90, longitude 0 to 359.75. It was taken from the public ARCO-ERA5 archive (`geopotential_at_surface`, a time-invariant field).
46
+
47
+ ### Outputs
48
+
49
+ Class index order (index 0 is background):
50
+
51
+ | Index | Class |
52
+ |---|---|
53
+ | 0 | background |
54
+ | 1 | cold front |
55
+ | 2 | warm front |
56
+ | 3 | stationary front |
57
+ | 4 | occluded front |
58
+ | 5 | dryline |
59
+ | 6 | trough |
60
+ | 7 | tropical trough |
61
+ | 8 | instability line |
62
+
63
+ The model was trained with an instability-line class ("no_inst" in its original filename), but that class is not used in the served product. The other seven classes are.
64
+
65
+ ### Usage
66
+
67
+ Load with Keras (`keras.saving.load_model`). Large domains are run in overlapping patches and stitched with blending. The serving pipeline (tiling, IFS ingest, derived variables, GeoZarr output) is described at fronts.espr.ai.
68
+
69
+ ## Attribution
70
+
71
+ Surface geopotential comes from ERA5 (Hersbach et al., ECMWF / Copernicus Climate Change Service). Please acknowledge ERA5 and ECMWF open data when using these inputs.
geopotential_at_surface_era5_0p25deg.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:595026b259654bab939733dfb3a74dd2cd79ac1134598cc90981e1b252e37b94
3
+ size 4153088
nbs_square_dilation_trof_tt_no_inst.keras → v2.0.0.keras RENAMED
File without changes