In [ ]:

# Uncomment to install climdata in Google Colab or other environments
# !pip install climdata

# Uncomment to install climdata in Google Colab or other environments # !pip install climdata

MSWX Dataset — climdata Example¶

MSWX (Multi-Source Weather) is a global, daily, 0.1° meteorological dataset based on the ERA5 reanalysis, bias-corrected against in-situ station observations. It is distributed via Google Drive and accessed in climdata through a Google Service Account.

Available variables:

Variable	Description	Units
tasmin	Daily minimum air temperature	°C
tasmax	Daily maximum air temperature	°C
tas	Daily mean air temperature	°C
pr	Daily precipitation	mm/day
rsds	Downward shortwave radiation	W/m²
hurs	Relative humidity	%
sfcWind	Wind speed	m/s

Prerequisites: A Google Service Account JSON key file with read access to the MSWX Google Drive folders is required for downloading data.

1. Setup & Imports¶

In [ ]:

from climdata import ClimData
from climdata.datasets.MSWX import MSWXmirror
import xarray as xr
import pandas as pd
import logging

logging.basicConfig(
    level=logging.INFO,
    format="%(levelname)s | %(message)s",
    force=True,
)

from climdata import ClimData from climdata.datasets.MSWX import MSWXmirror import xarray as xr import pandas as pd import logging logging.basicConfig( level=logging.INFO, format="%(levelname)s | %(message)s", force=True, )

2. Configuration¶

Set up the configuration overrides. The google_service_account path must point to your Google Service Account JSON key.

In [ ]:

overrides = [
    "dataset=mswx",                             # Select the MSWX dataset
    "region=germany",                           # Predefined bounding box
    "time_range.start_date=2010-01-01",         # Start date for data extraction
    "time_range.end_date=2010-03-31",           # End date for data extraction
    "variables=[tasmin,tasmax,pr]",             # Variables to extract
    "data_dir=./data",                          # Local directory to store downloaded files
    "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json",  # Service account key
]

extractor = ClimData(overrides=overrides)

overrides = [ "dataset=mswx", # Select the MSWX dataset "region=germany", # Predefined bounding box "time_range.start_date=2010-01-01", # Start date for data extraction "time_range.end_date=2010-03-31", # End date for data extraction "variables=[tasmin,tasmax,pr]", # Variables to extract "data_dir=./data", # Local directory to store downloaded files "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json", # Service account key ] extractor = ClimData(overrides=overrides)

3. Download Only¶

fetch() is the download step on its own: for each variable it works out the expected daily filenames over time_range, skips the ones already on disk, and pulls only the missing ones from Google Drive. It returns the list of local filenames — no NetCDF file is opened and nothing is read into memory.

Use this when you just want to stage the files locally (e.g. on a login node, or ahead of a batch job) and read them later.

In [ ]:

# Download only — no reading, no xarray.
cfg = extractor.cfg
mswx_dl = MSWXmirror(cfg)
mswx_info = cfg.dsinfo[cfg.dataset.upper()]

for var in cfg.variables:
    folder_id = mswx_info["variables"][var]["folder_id"]
    files = mswx_dl.fetch(folder_id, var)
    print(f"{var}: {len(files)} files staged in {cfg.data_dir}/{cfg.dataset.upper()}/{var.upper()}")

# Download only — no reading, no xarray. cfg = extractor.cfg mswx_dl = MSWXmirror(cfg) mswx_info = cfg.dsinfo[cfg.dataset.upper()] for var in cfg.variables: folder_id = mswx_info["variables"][var]["folder_id"] files = mswx_dl.fetch(folder_id, var) print(f"{var}: {len(files)} files staged in {cfg.data_dir}/{cfg.dataset.upper()}/{var.upper()}")

4. Download & Load Data¶

extract() runs the same fetch() download above and then reads every file into a single xarray.Dataset. If you already ran the download-only cell, nothing is re-downloaded here — only the read happens.

In [ ]:

ds = extractor.extract()
print(ds)

ds = extractor.extract() print(ds)

5. Inspect the Dataset¶

In [ ]:

# Inspect one variable
ds['tasmin']

# Inspect one variable ds['tasmin']

In [ ]:

# Convert to pandas DataFrame for quick inspection
df_tasmin = ds['tasmin'].to_dataframe().reset_index()
df_tasmin.head()

# Convert to pandas DataFrame for quick inspection df_tasmin = ds['tasmin'].to_dataframe().reset_index() df_tasmin.head()

6. Spatial Extraction — Bounding Box¶

In [ ]:

overrides_box = [
    "dataset=mswx",
    "time_range.start_date=2010-01-01",
    "time_range.end_date=2010-03-31",
    "variables=[tasmax,pr]",
    "data_dir=./data",
    "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json",
]

extractor_box = ClimData(overrides=overrides_box)
mswx_box = MSWXmirror(extractor_box.cfg)

# Extract a bounding box (Germany)
mswx_box.extract(box={
    "lon_min": 5.8,
    "lon_max": 15.0,
    "lat_min": 47.3,
    "lat_max": 55.1
})

ds_box = mswx_box.load(variable="tasmax")
print(ds_box)

overrides_box = [ "dataset=mswx", "time_range.start_date=2010-01-01", "time_range.end_date=2010-03-31", "variables=[tasmax,pr]", "data_dir=./data", "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json", ] extractor_box = ClimData(overrides=overrides_box) mswx_box = MSWXmirror(extractor_box.cfg) # Extract a bounding box (Germany) mswx_box.extract(box={ "lon_min": 5.8, "lon_max": 15.0, "lat_min": 47.3, "lat_max": 55.1 }) ds_box = mswx_box.load(variable="tasmax") print(ds_box)

7. Run the Full Workflow via ClimData¶

Use ClimData.run_workflow() to chain extraction, optional imputation, climate index calculation, and saving to NetCDF in one call.

In [ ]:

overrides_wf = [
    "dataset=mswx",
    "lat=52",
    "lon=13",
    "time_range.start_date=2010-01-01",
    "time_range.end_date=2010-12-31",
    "variables=[tasmin,tasmax,pr]",
    "data_dir=./data",
    "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json",
    "index=tn10p",      # Cold nights climate extreme index
    "impute=BRITS",     # Deep-learning gap-filling (optional)
]

seq = ["extract", "impute", "calc_index", "to_nc"]

extractor_wf = ClimData(overrides=overrides_wf)
result = extractor_wf.run_workflow(actions=seq)
result

overrides_wf = [ "dataset=mswx", "lat=52", "lon=13", "time_range.start_date=2010-01-01", "time_range.end_date=2010-12-31", "variables=[tasmin,tasmax,pr]", "data_dir=./data", "dsinfo.mswx.params.google_service_account=./.climdata_conf/service.json", "index=tn10p", # Cold nights climate extreme index "impute=BRITS", # Deep-learning gap-filling (optional) ] seq = ["extract", "impute", "calc_index", "to_nc"] extractor_wf = ClimData(overrides=overrides_wf) result = extractor_wf.run_workflow(actions=seq) result

8. Save to CSV¶

Write the extracted data out as a flat CSV file.

In [ ]:

mswx_box.save_csv("mswx_tasmax_2010.csv")
print("Saved to mswx_tasmax_2010.csv")

mswx_box.save_csv("mswx_tasmax_2010.csv") print("Saved to mswx_tasmax_2010.csv")