42. Python and Jupyter Notebook¶

Download a NetCDF subset of IMERG V07 daily satellite rainfall, then inspect it with xarray. This example uses a small bounding box and fixed historical dates. For a forecast, change the product, fields and dates together using the catalogue.

42.1 Install and authenticate¶

pip install requests xarray netCDF4 notebook
export GAP_API_TOKEN='YOUR_API_KEY'

Create a key using the API key guide. Launch Jupyter from the same environment so the notebook can read GAP_API_TOKEN.

42.2 Run the example¶

Download sample.py or sample.ipynb. Run python sample.py, or open the notebook with jupyter notebook and run its cell.

The script:

1. Requests precipitation from imerg_v07 for 1–3 September 2026.
2. Downloads into a temporary file and checks the HTTP status before replacing data.nc.
3. Opens the completed file with xarray and prints coordinates, units and variables.

Errors raise an exception. If a request fails, an older data.nc is not reopened as though it were a new result.

42.3 Change the request¶

params = {
    "product": "imerg_v07",
    "attributes": "precipitation",
    "start_date": "2026-09-01",
    "end_date": "2026-09-03",
    "output_type": "netcdf",
    "bbox": "36.7,-1.4,36.9,-1.2",
}

A bounding box is west, south, east, north. For a point, remove bbox and supply both lat and lon. JSON supports a single point; adapt the response handling to call response.json() rather than saving it as NetCDF.

For NextGen daily forecasts, use nextgen_forecast with fields such as precip_accum_p50,precip_accum_p75,tmin_c,tmax_c and available forecast dates. Read run selection before pinning forecast_date.

42.4 Inspect the data¶

import xarray as xr

with xr.open_dataset("data.nc") as dataset:
    print(dataset)
    print(dataset["precipitation"].attrs)
    print(dataset.coords)

Dimension names and shapes depend on the product; do not assume a fixed grid. Preserve missing values. For long requests use background jobs.