Training on CAMS data in forecasting mode

config/config_forecasting.yml

@@ -141,7 +141,7 @@ training_config:
   shuffle: True
 
   start_date: 1979-01-01T00:00
-  end_date: 2022-12-31T00:00
+  end_date: 2021-05-31T12:00
 
   time_window_step: 06:00:00
   time_window_len: 06:00:00
@@ -195,8 +195,8 @@ validation_config:
   samples_per_mini_epoch: 256
   shuffle: False
 
-  start_date: 2023-10-01T00:00
-  end_date: 2023-12-31T00:00
+  start_date: 2021-06-01T12:00
+  end_date: 2022-05-31T12:00
 
   # whether to track the exponential moving average of weights for validation
   validate_with_ema: 

config/evaluate/eval_config.yml

@@ -78,6 +78,30 @@ run_ids :
     label: "pretrained model ar40mckx"
     results_base_dir : "./results/"
     #NEW: if "streams" is not specified, the default streams are used
+    # CAMS comparison (optional): remove the comment markers below to enable
+    # cams_comparison:
+    #   cams_base_dir: "/path/to/cams"                      # required
+    #   cams_forecast_filename: "cams_forecast_2022.zarr"   # default shown
+    #   cams_stream: "CAMSEAC4"                             # stream name used for output paths
+    #   cams_channels:                                      # null => auto-detect from CAMS store
+    #     - "co_500"
+    #     - "no2_850"
+    #     - "o3_500"
+    #     - "tc_co"
+    #     - "pm2p5"
+    #   forecast_steps: "all"                               # "all", int, or list of hours
+    #   step_hrs: 6                                         # WG time-step in hours (optional)
+    #   plot_bias_maps: true
+    #   plot_value_maps: false
+    #   plot_rmse_curves: true
+    #   write_scorecard: true
+    #   create_video: false
+    #   fps: 2.0
+    #   convert_to_ppm: false                               # convert kg/kg profiles to ppmv
+    #   color_max_ppb: null                                 # symmetric colour-scale cap in ppb
+    #   rmse_rel_pct_clip: 40.0                             # clip relative RMSE% in scorecard heatmap
+    #   regrid_forecast: false
+    #   cams_analysis_filename: null                        # optional: zarr file for native-grid RMSE
 
 
   c8g5katp:

config/streams/era5_1deg_cams_eac4_forecasting/cams_eac4.yml

@@ -0,0 +1,66 @@
+# (C) Copyright 2024 WeatherGenerator contributors.
+#
+# This software is licensed under the terms of the Apache Licence Version 2.0
+# which can be obtained at http://www.apache.org/licenses/LICENSE-2.0.
+#
+# In applying this licence, ECMWF does not waive the privileges and immunities
+# granted to it by virtue of its status as an intergovernmental organisation
+# nor does it submit to any jurisdiction.
+
+CAMSEAC4 :
+  type : cams
+  filenames : ['cams_eac4_2003_2024.zarr']
+  stream_id : 1
+  source : ['no_925', 'no_500', 'no2_925', 'no2_500', 'co_925', 'co_500' ]
+  target : ['no_925', 'no_500', 'no2_925', 'no2_500', 'co_925', 'co_500' ]
+  source_exclude : []
+  target_exclude : []
+  variables: [
+      # Surface variables
+      'pm1', 'pm2p5', 'pm10',
+      'tc_co', 'tc_no', 'tc_no2', 'tc_o3', 'tc_so2',
+
+      # Ozone (o3)
+      'go3_1000', 'go3_925', 'go3_850', 'go3_700', 'go3_600', 'go3_500',
+      'go3_400', 'go3_300', 'go3_250', 'go3_200', 'go3_150', 'go3_100', 'go3_50',
+
+      # Sulfur dioxide (so2)
+      'so2_1000', 'so2_925', 'so2_850', 'so2_700', 'so2_600', 'so2_500',
+      'so2_400', 'so2_300', 'so2_250', 'so2_200', 'so2_150', 'so2_100', 'so2_50',
+
+      # Nitrogen monoxide (no)
+      'no_1000', 'no_925', 'no_850', 'no_700', 'no_600', 'no_500',
+      'no_400', 'no_300', 'no_250', 'no_200', 'no_150', 'no_100', 'no_50',
+
+      # Nitrogen dioxide (no2)
+      'no2_1000', 'no2_925', 'no2_850', 'no2_700', 'no2_600', 'no2_500',
+      'no2_400', 'no2_300', 'no2_250', 'no2_200', 'no2_150', 'no2_100', 'no2_50',
+
+      # Carbon monoxide (co)
+      'co_1000', 'co_925', 'co_850', 'co_700', 'co_600', 'co_500',
+      'co_400', 'co_300', 'co_250', 'co_200', 'co_150', 'co_100', 'co_50',
+  ]
+  pressure_levels : ["50", "100", "150", "200", "250", "300", "400", "500", "600", "700", "850", "925", "1000"]
+  loss_weight : 1.
+  masking_rate : 0.6
+  masking_rate_none : 0.05
+  token_size : 16
+  tokenize_spacetime : True
+  max_num_targets: -1
+  embed : 
+    net : transformer
+    num_tokens : 1
+    num_heads : 8
+    dim_embed : 256 # 512
+    num_blocks : 2
+  embed_target_coords :
+    net : linear
+    dim_embed : 256 # 512
+  target_readout :
+    type : 'obs_value'  # token or obs_value
+    num_layers : 2
+    num_heads : 4
+    # sampling_rate : 0.2
+  pred_head :
+    ens_size : 1
+    num_layers : 1

config/streams/era5_1deg_cams_eac4_forecasting/era5.yml

@@ -0,0 +1,104 @@
+# (C) Copyright 2024 WeatherGenerator contributors.
+#
+# This software is licensed under the terms of the Apache Licence Version 2.0
+# which can be obtained at http://www.apache.org/licenses/LICENSE-2.0.
+#
+# In applying this licence, ECMWF does not waive the privileges and immunities
+# granted to it by virtue of its status as an intergovernmental organisation
+# nor does it submit to any jurisdiction.
+
+ERA5 :
+  type : anemoi
+  filenames : ['aifs-ea-an-oper-0001-mars-o96-1979-2022-6h-v6.zarr']
+  stream_id : 0
+  source_exclude : ['w_', 'skt', 'tcw', 'cp', 'tp']
+  target_exclude : ['w_', 'slor', 'sdor', 'tcw', 'cp', 'tp']
+  loss_weight : 1.
+  location_weight : cosine_latitude
+  masking_rate : 0.6
+  masking_rate_none : 0.05
+  token_size : 8
+  tokenize_spacetime : True
+  max_num_targets: 20000
+  embed : 
+    net : transformer
+    num_tokens : 1
+    num_heads : 8
+    dim_embed : 256
+    num_blocks : 2
+  embed_target_coords :
+    net : linear
+    dim_embed : 256
+  target_readout :
+    num_layers : 2
+    num_heads : 4
+    # sampling_rate : 0.2
+  pred_head :
+    ens_size : 1
+    num_layers : 1
+  channel_weights :
+    q_50: 0.2
+    q_100: 0.23
+    q_150: 0.26
+    q_200: 0.29
+    q_250: 0.33
+    q_300: 0.36
+    q_400: 0.42
+    q_500: 0.48
+    q_600: 0.55
+    q_700: 0.61
+    q_850: 0.71
+    q_925: 0.75
+    q_1000: 0.8
+    t_50: 0.2
+    t_100: 0.23
+    t_150: 0.26
+    t_200: 0.29
+    t_250: 0.33
+    t_300: 0.36
+    t_400: 0.42
+    t_500: 0.48
+    t_600: 0.55
+    t_700: 0.61
+    t_850: 0.71
+    t_925: 0.75
+    t_1000: 0.8
+    u_50: 0.2
+    u_100: 0.23
+    u_150: 0.26
+    u_200: 0.29
+    u_250: 0.33
+    u_300: 0.36
+    u_400: 0.42
+    u_500: 0.48
+    u_600: 0.55
+    u_700: 0.61
+    u_850: 0.71
+    u_925: 0.75
+    u_1000: 0.8
+    v_50: 0.2
+    v_100: 0.23
+    v_150: 0.26
+    v_200: 0.29
+    v_250: 0.33
+    v_300: 0.36
+    v_400: 0.42
+    v_500: 0.48
+    v_600: 0.55
+    v_700: 0.61
+    v_850: 0.71
+    v_925: 0.75
+    v_1000: 0.8
+    z_50: 0.2
+    z_100: 0.23
+    z_150: 0.26
+    z_200: 0.29
+    z_250: 0.33
+    z_300: 0.36
+    z_400: 0.42
+    z_500: 0.48
+    z_600: 0.55
+    z_700: 0.61
+    z_850: 0.71
+    z_925: 0.75
+    z_1000: 0.8