rguilcas commited on
Commit
516f61b
·
verified ·
1 Parent(s): 0e297c4

Faster app v2 with caching

Browse files
Files changed (1) hide show
  1. app.py +75 -56
app.py CHANGED
@@ -20,7 +20,7 @@ active_tab_css = """
20
  pn.extension(design="material", sizing_mode="stretch_width", raw_css=[active_tab_css])
21
 
22
 
23
- # @pn.cache # Temporarily disabled to reload with new precursor files
24
  def get_data():
25
  ds1= xr.open_dataset("data/LENS_precomputed_terms_combined.nc")
26
  gdf_regions = gpd.read_file('data/rainfall_regions.geojson',).set_index('index')
@@ -46,8 +46,18 @@ def get_data():
46
  ds1, gdf_regions, gdf_precursors_u850, gdf_precursors_v850, gdf_precursors_z500 = get_data()
47
 
48
 
49
- def plot_precursor_map_gdf(gdf, variable_name, season='SON', region='Egypt + East Libya', vmax=1000):
50
- """Plot a single precursor map from a GeoDataFrame."""
 
 
 
 
 
 
 
 
 
 
51
  filtered = gdf.query("season==@season & region==@region")
52
  if len(filtered) == 0:
53
  return pn.pane.Markdown(f"No data for {region} in {season}")
@@ -70,9 +80,9 @@ def plot_precursor_map_gdf(gdf, variable_name, season='SON', region='Egypt + Eas
70
 
71
  def get_precursor_plot(season='SON', region='Egypt + East Libya'):
72
  """Get all three precursor maps side-by-side."""
73
- u850_map = plot_precursor_map_gdf(gdf_precursors_u850, 'U850', vmax=10, region=region, season=season)
74
- v850_map = plot_precursor_map_gdf(gdf_precursors_v850, 'V850', vmax=10, region=region, season=season)
75
- z500_map = plot_precursor_map_gdf(gdf_precursors_z500, 'Z500', vmax=2000, region=region, season=season)
76
 
77
  return (u850_map + v850_map + z500_map).cols(3)
78
 
@@ -80,6 +90,62 @@ def get_precursor_plot(season='SON', region='Egypt + East Libya'):
80
 
81
 
82
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
83
  class CMIPApp(param.Parameterized):
84
  model = param.Selector(default='CESM2', objects=list(ds1.model.values))
85
  season = param.Selector(default='DJF', objects=list(ds1.season.values))
@@ -156,56 +222,9 @@ class CMIPApp(param.Parameterized):
156
 
157
  def _get_classified_data(self):
158
  """Helper to unify classification and color mapping for all regions."""
159
- # 1. Select data components for all regions
160
- ds_model = ds1.sel(model=self.model, season=self.season, term=self.term)
161
- # Sum over synoptic bins and correct units
162
- ds_model = (ds_model.sum("synoptic_bin") / 0.05 * 100).individual_term
163
-
164
- # 2. Extract base components
165
- conversion = ds_model.sel(source='Conversion')
166
- dynamical = ds_model.sel(source='Dynamical')
167
- nonlinear = ds_model.sel(source='Non Linear')
168
-
169
- # 3. Determine X and Y for classification (based on nonlinear_pos setting)
170
- if self.nonlinear_pos == 'Dynamical':
171
- x_val = conversion
172
- y_val = dynamical + nonlinear
173
- else:
174
- x_val = conversion + nonlinear
175
- y_val = dynamical
176
-
177
- # 4. Create DataFrame
178
- df = pd.DataFrame({
179
- 'Region': ds_model.region_id.values,
180
- 'Conversion': conversion.values,
181
- 'Dynamical': dynamical.values,
182
- 'Non_Linear': nonlinear.values,
183
- 'X': x_val.values,
184
- 'Y': y_val.values,
185
- 'Selected_Sum': ds_model.sel(source=self.source).sum('source').values
186
- })
187
-
188
- # 5. Add classification logic
189
- def classify(row):
190
- x, y = row['X'], row['Y']
191
- if abs(x) + abs(y) <= 20: return 'Minimal'
192
- if abs(x) < 0.2 * abs(y): return 'Dynamical'
193
- if abs(y) < 0.2 * abs(x): return 'Conversion'
194
- return 'Compounding' if x * y > 0 else 'Cancelling'
195
-
196
- df['Type'] = df.apply(classify, axis=1)
197
-
198
- # 6. Saturated colors for readable hover text
199
- text_cmap = {
200
- 'Minimal': 'grey',
201
- 'Conversion': '#b8860b', # DarkGoldenRod
202
- 'Dynamical': 'blue',
203
- 'Compounding': 'red',
204
- 'Cancelling': 'green'
205
- }
206
- df['Text_Color'] = df['Type'].map(text_cmap)
207
-
208
- return df
209
 
210
  @param.depends('model', 'season', 'term', 'source', 'vmax', '_selection_index')
211
  def get_value_map(self):
 
20
  pn.extension(design="material", sizing_mode="stretch_width", raw_css=[active_tab_css])
21
 
22
 
23
+ @pn.cache()
24
  def get_data():
25
  ds1= xr.open_dataset("data/LENS_precomputed_terms_combined.nc")
26
  gdf_regions = gpd.read_file('data/rainfall_regions.geojson',).set_index('index')
 
46
  ds1, gdf_regions, gdf_precursors_u850, gdf_precursors_v850, gdf_precursors_z500 = get_data()
47
 
48
 
49
+ @pn.cache(max_items=128)
50
+ def get_cached_precursor_map(variable_name, season, region, vmax):
51
+ """Cached version of precursor map plotting."""
52
+ if variable_name == 'U850':
53
+ gdf = gdf_precursors_u850
54
+ elif variable_name == 'V850':
55
+ gdf = gdf_precursors_v850
56
+ elif variable_name == 'Z500':
57
+ gdf = gdf_precursors_z500
58
+ else:
59
+ return None
60
+
61
  filtered = gdf.query("season==@season & region==@region")
62
  if len(filtered) == 0:
63
  return pn.pane.Markdown(f"No data for {region} in {season}")
 
80
 
81
  def get_precursor_plot(season='SON', region='Egypt + East Libya'):
82
  """Get all three precursor maps side-by-side."""
83
+ u850_map = get_cached_precursor_map('U850', season, region, 10)
84
+ v850_map = get_cached_precursor_map('V850', season, region, 10)
85
+ z500_map = get_cached_precursor_map('Z500', season, region, 2000)
86
 
87
  return (u850_map + v850_map + z500_map).cols(3)
88
 
 
90
 
91
 
92
 
93
+ @pn.cache(max_items=32)
94
+ def compute_classified_data(model, season, term, nonlinear_pos, source_tuple):
95
+ """Cached computation of classified data."""
96
+ # 1. Select data components for all regions
97
+ ds_model = ds1.sel(model=model, season=season, term=term)
98
+ # Sum over synoptic bins and correct units
99
+ ds_model = (ds_model.sum("synoptic_bin") / 0.05 * 100).individual_term
100
+
101
+ # 2. Extract base components
102
+ conversion = ds_model.sel(source='Conversion')
103
+ dynamical = ds_model.sel(source='Dynamical')
104
+ nonlinear = ds_model.sel(source='Non Linear')
105
+
106
+ # 3. Determine X and Y for classification (based on nonlinear_pos setting)
107
+ if nonlinear_pos == 'Dynamical':
108
+ x_val = conversion
109
+ y_val = dynamical + nonlinear
110
+ else:
111
+ x_val = conversion + nonlinear
112
+ y_val = dynamical
113
+
114
+ # 4. Create DataFrame
115
+ # ds_model.sel(source=list(source_tuple)) works because xarray handles lists
116
+ df = pd.DataFrame({
117
+ 'Region': ds_model.region_id.values,
118
+ 'Conversion': conversion.values,
119
+ 'Dynamical': dynamical.values,
120
+ 'Non_Linear': nonlinear.values,
121
+ 'X': x_val.values,
122
+ 'Y': y_val.values,
123
+ 'Selected_Sum': ds_model.sel(source=list(source_tuple)).sum('source').values
124
+ })
125
+
126
+ # 5. Add classification logic
127
+ def classify(row):
128
+ x, y = row['X'], row['Y']
129
+ if abs(x) + abs(y) <= 20: return 'Minimal'
130
+ if abs(x) < 0.2 * abs(y): return 'Dynamical'
131
+ if abs(y) < 0.2 * abs(x): return 'Conversion'
132
+ return 'Compounding' if x * y > 0 else 'Cancelling'
133
+
134
+ df['Type'] = df.apply(classify, axis=1)
135
+
136
+ # 6. Saturated colors for readable hover text
137
+ text_cmap = {
138
+ 'Minimal': 'grey',
139
+ 'Conversion': '#b8860b', # DarkGoldenRod
140
+ 'Dynamical': 'blue',
141
+ 'Compounding': 'red',
142
+ 'Cancelling': 'green'
143
+ }
144
+ df['Text_Color'] = df['Type'].map(text_cmap)
145
+
146
+ return df
147
+
148
+
149
  class CMIPApp(param.Parameterized):
150
  model = param.Selector(default='CESM2', objects=list(ds1.model.values))
151
  season = param.Selector(default='DJF', objects=list(ds1.season.values))
 
222
 
223
  def _get_classified_data(self):
224
  """Helper to unify classification and color mapping for all regions."""
225
+ # Convert list to tuple for hashing
226
+ source_tuple = tuple(self.source) if self.source else ()
227
+ return compute_classified_data(self.model, self.season, self.term, self.nonlinear_pos, source_tuple)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
228
 
229
  @param.depends('model', 'season', 'term', 'source', 'vmax', '_selection_index')
230
  def get_value_map(self):