microsoft · rejuvyesh · Sep 30, 2023 · Sep 18, 2023 · Sep 29, 2023 · Sep 29, 2023
diff --git a/.gitignore b/.gitignore
@@ -133,3 +133,6 @@ dmypy.json
 
 # experiments
 exps
+
+# snakemake logs
+.snakemake
diff --git a/docs/usage.md b/docs/usage.md
@@ -4,7 +4,21 @@
 
 ### Data Preparation
 
-The code for downloading and preprocessing CMIP6 data is coming soon
+First install `snakemake` following [these instructions](https://snakemake.readthedocs.io/en/stable/getting_started/installation.html)
+
+To download and regrid a CMIP6 dataset to a common resolution (e.g., 1.406525 degree), go to the corresponding directory inside `snakemake_configs` and run
+```bash
+snakemake all --configfile config_2m_temperature.yml --cores 8
+```
+This script will download and regrid the `2m_temperature` data in parallel using 8 CPU cores. Modify `configfile` for other variables. After downloading and regrdding, run the following script to preprocess the `.nc` files into `.npz` format for pretraining ClimaX
+```bash
+python src/data_preprocessing/nc2np_equally_cmip6.py \
+    --dataset mpi
+    --path /data/CMIP6/MPI-ESM/1.40625deg/
+    --num_shards 10
+    --save_dir /data/CMIP6/MPI-ESM/1.40625deg_np_10shards
+```
+in which `num_shards` denotes the number of chunks to break each `.nc` file into.
 
 ### Training
 

diff --git a/snakemake_configs/AWI-ESM/Snakefile b/snakemake_configs/AWI-ESM/Snakefile
@@ -0,0 +1,49 @@
+
+year_strings = [f'{y}01010600-{y+1}01010000' for y in range(1850, 2015, 1)]
+
+print(config)
+
+rule download:
+    output:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc",
+    shell:
+         "wget https://esgf-data1.llnl.gov/thredds/fileServer/css03_data/CMIP6/CMIP/AWI/AWI-ESM-1-1-LR/historical/{config["
+         "run]}/6hrPlevPt/"
+         "{config[cmip_name]}/gn/v20200212/"
+         "{config[cmip_name]}_6hrPlevPt_AWI-ESM-1-1-LR_historical_{config[run]}_gn_{wildcards.year_str}.nc "
+         "-O {wildcards.dataset}/raw/{config[name]}/{config[name]}_{wildcards.year_str}_raw.nc"
+
+rule regrid:
+    input:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc"
+    output:
+          "{dataset}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc.tmp"
+    shell:
+          "python ../../src/data_preprocessing/regrid.py \
+            --input_fns {input} \
+            --output_dir {wildcards.dataset}/{wildcards.res}deg/{wildcards.name} \
+            --ddeg_out {wildcards.res} \
+            --cmip 1 \
+            --rename {config[cmip_name]} {config[era_name]} \
+            --file_ending nc.tmp"
+
+rule delete:
+    input:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc.tmp",
+                 res=config['res']),
+    output:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc",
+                 res=config['res'])
+    priority: 100
+    run:
+          for i, o in zip(input, output):
+              shell("mv {i} {o}")
+      #     shell("rm {wildcards.dataset}/raw/{wildcards.name}/{wildcards.name}_{wildcards.year_str}_raw.nc"),
+
+
+rule all:
+    input:
+         expand("{datadir}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc",
+                datadir=config['datadir'], res=config['res'], name=config['name'], year_str=year_strings)
+
+
diff --git a/snakemake_configs/AWI-ESM/config_10m_u_component_of_wind.yml b/snakemake_configs/AWI-ESM/config_10m_u_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: 10m_u_component_of_wind
+cmip_name: uas
+era_name: u10
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_10m_v_component_of_wind.yml b/snakemake_configs/AWI-ESM/config_10m_v_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: 10m_v_component_of_wind
+cmip_name: vas
+era_name: v10
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_2m_temperature.yml b/snakemake_configs/AWI-ESM/config_2m_temperature.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: 2m_temperature
+cmip_name: tas
+era_name: t2m
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_geopotential.yml b/snakemake_configs/AWI-ESM/config_geopotential.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: geopotential
+cmip_name: zg
+era_name: z
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_specific_humidity.yml b/snakemake_configs/AWI-ESM/config_specific_humidity.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: specific_humidity
+cmip_name: hus
+era_name: q
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_temperature.yml b/snakemake_configs/AWI-ESM/config_temperature.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: temperature
+cmip_name: ta
+era_name: t
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_u_component_of_wind.yml b/snakemake_configs/AWI-ESM/config_u_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: u_component_of_wind
+cmip_name: ua
+era_name: u
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/AWI-ESM/config_v_component_of_wind.yml b/snakemake_configs/AWI-ESM/config_v_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/AWI-ESM
+name: v_component_of_wind
+cmip_name: va
+era_name: v
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/CMCC/Snakefile b/snakemake_configs/CMCC/Snakefile
@@ -0,0 +1,50 @@
+year_strings = [f'{y}01010600-{y+1}01010000' for y in range(1850, 2015, 1)]
+
+print(config)
+
+rule download:
+    output:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc",
+    shell:
+         "wget https://esgf.ceda.ac.uk/thredds/fileServer/esg_cmip6/CMIP6/CMIP/CMCC/CMCC-CM2-HR4/historical/{config["
+         "run]}/6hrPlevPt/"
+         "{config[cmip_name]}/gn/v20200904/"
+         "{config[cmip_name]}_6hrPlevPt_CMCC-CM2-HR4_historical_{config[run]}_gn_{wildcards.year_str}.nc "
+         "-O {wildcards.dataset}/raw/{config[name]}/{config[name]}_{wildcards.year_str}_raw.nc"
+
+      #    https://esgf.ceda.ac.uk/thredds/fileServer/esg_cmip6/CMIP6/CMIP/CMCC/CMCC-CM2-HR4/historical/r1i1p1f1/6hrPlevPt/ta/gn/v20200904/ta_6hrPlevPt_CMCC-CM2-HR4_historical_r1i1p1f1_gn_185001010600-185101010000.nc
+
+rule regrid:
+    input:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc"
+    output:
+          "{dataset}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc.tmp"
+    shell:
+          "python ../../src/data_preprocessing/regrid.py \
+            --input_fns {input} \
+            --output_dir {wildcards.dataset}/{wildcards.res}deg/{wildcards.name} \
+            --ddeg_out {wildcards.res} \
+            --cmip 1 \
+            --rename {config[cmip_name]} {config[era_name]} \
+            --file_ending nc.tmp"
+
+rule delete:
+    input:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc.tmp",
+                 res=config['res']),
+    output:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc",
+                 res=config['res'])
+    priority: 100
+    run:
+          for i, o in zip(input, output):
+              shell("mv {i} {o}")
+      #     shell("rm {wildcards.dataset}/raw/{wildcards.name}/{wildcards.name}_{wildcards.year_str}_raw.nc"),
+
+
+rule all:
+    input:
+         expand("{datadir}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc",
+                datadir=config['datadir'], res=config['res'], name=config['name'], year_str=year_strings)
+
+
diff --git a/snakemake_configs/CMCC/config_geopotential.yml b/snakemake_configs/CMCC/config_geopotential.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/CMCC
+name: geopotential
+cmip_name: zg
+era_name: z
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/CMCC/config_temperature.yml b/snakemake_configs/CMCC/config_temperature.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/CMCC
+name: temperature
+cmip_name: ta
+era_name: t
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/CMCC/config_u_component_of_wind.yml b/snakemake_configs/CMCC/config_u_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/CMCC
+name: u_component_of_wind
+cmip_name: ua
+era_name: u
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/CMCC/config_v_component_of_wind.yml b/snakemake_configs/CMCC/config_v_component_of_wind.yml
@@ -0,0 +1,8 @@
+datadir: /data/CMIP6/CMCC
+name: v_component_of_wind
+cmip_name: va
+era_name: v
+run: r1i1p1f1
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/Snakefile b/snakemake_configs/HAMMOZ/Snakefile
@@ -0,0 +1,59 @@
+
+year_strings = [
+    '185001010600-187001010000',
+    '187001010600-189001010000',
+    '189001010600-191001010000',
+    '191001010600-193001010000',
+    '193001010600-195001010000',
+    '195001010600-197001010000',
+    '197001010600-199001010000',
+    '199001010600-201001010000',
+    '201001010600-201501010000',
+]
+
+print(config)
+
+rule download:
+    output:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc",
+    shell:
+         "wget https://esgf-data1.llnl.gov/thredds/fileServer/css03_data/CMIP6/CMIP/HAMMOZ-Consortium/MPI-ESM-1-2-HAM/historical/{config["
+         "run]}/6hrPlevPt/"
+         "{config[cmip_name]}/gn/{config[version]}/"
+         "{config[cmip_name]}_6hrPlevPt_MPI-ESM-1-2-HAM_historical_{config[run]}_gn_{wildcards.year_str}.nc "
+         "-O {wildcards.dataset}/raw/{config[name]}/{config[name]}_{wildcards.year_str}_raw.nc"
+
+rule regrid:
+    input:
+          "{dataset}/raw/{name}/{name}_{year_str}_raw.nc"
+    output:
+          "{dataset}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc.tmp"
+    shell:
+          "python ../../src/data_preprocessing/regrid.py \
+            --input_fns {input} \
+            --output_dir {wildcards.dataset}/{wildcards.res}deg/{wildcards.name} \
+            --ddeg_out {wildcards.res} \
+            --cmip 1 \
+            --rename {config[cmip_name]} {config[era_name]} \
+            --file_ending nc.tmp"
+
+rule delete:
+    input:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc.tmp",
+                 res=config['res']),
+    output:
+          expand("{{dataset}}/{res}deg/{{name}}/{{name}}_{{year_str}}_{res}deg.nc",
+                 res=config['res'])
+    priority: 100
+    run:
+          for i, o in zip(input, output):
+              shell("mv {i} {o}")
+      #     shell("rm {wildcards.dataset}/raw/{wildcards.name}/{wildcards.name}_{wildcards.year_str}_raw.nc"),
+
+
+rule all:
+    input:
+         expand("{datadir}/{res}deg/{name}/{name}_{year_str}_{res}deg.nc",
+                datadir=config['datadir'], res=config['res'], name=config['name'], year_str=year_strings)
+
+
diff --git a/snakemake_configs/HAMMOZ/config_10m_u_component_of_wind.yml b/snakemake_configs/HAMMOZ/config_10m_u_component_of_wind.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: 10m_u_component_of_wind
+cmip_name: uas
+era_name: u10
+run: r1i1p1f1
+version: v20190627
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_10m_v_component_of_wind.yml b/snakemake_configs/HAMMOZ/config_10m_v_component_of_wind.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: 10m_v_component_of_wind
+cmip_name: vas
+era_name: v10
+run: r1i1p1f1
+version: v20190627
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_2m_temperature.yml b/snakemake_configs/HAMMOZ/config_2m_temperature.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: 2m_temperature
+cmip_name: tas
+era_name: t2m
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_geopotential.yml b/snakemake_configs/HAMMOZ/config_geopotential.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: geopotential
+cmip_name: zg
+era_name: z
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_specific_humidity.yml b/snakemake_configs/HAMMOZ/config_specific_humidity.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: specific_humidity
+cmip_name: hus
+era_name: q
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_temperature.yml b/snakemake_configs/HAMMOZ/config_temperature.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: temperature
+cmip_name: ta
+era_name: t
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_u_component_of_wind.yml b/snakemake_configs/HAMMOZ/config_u_component_of_wind.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: u_component_of_wind
+cmip_name: ua
+era_name: u
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625
diff --git a/snakemake_configs/HAMMOZ/config_v_component_of_wind.yml b/snakemake_configs/HAMMOZ/config_v_component_of_wind.yml
@@ -0,0 +1,9 @@
+datadir: /data/CMIP6/HAMMOZ
+name: v_component_of_wind
+cmip_name: va
+era_name: v
+run: r1i1p1f1
+version: v20190628
+res:
+  - 1.40625
+  # - 5.625