Era5-download

[ReisQuarteu]

I am trying to use data_ingestion/weather/download_era5 workflow. I filtered region of interest and time range but the workflow ignored the 'region of interest' and returned data from all over the world and for the first selected year (2021).
The code that I am using is next presented. What is the problem? I thank you in advance.

from datetime import datetime
from shapely import wkt
from vibe_core.client import get_default_vibe_client
WORKFLOW_NAME = "data_ingestion/weather/download_era5"
RUN_NAME = "era5 download"
input_geometry_path = "./input_region.wkt"
time_range = (datetime(2020,1,1), datetime(2023, 12, 31))

# Reading the geometry file
with open(input_geometry_path) as f:
    geometry = wkt.load(f)

client = get_default_vibe_client()
client.document_workflow(WORKFLOW_NAME)
wf_run = client.run(WORKFLOW_NAME, RUN_NAME, geometry=geometry, time_range=time_range)

and the file input_geometry_path.wkt has the following content:

POLYGON((-7.4707 41.7037,-7.4583 41.2181, -6.7044 41.2264,-6.7072 41.7068,-7.4707 41.7037))

[rafaspadilha]

Hi, @ReisQuarteu. Thank you for raising the question.

The output ERA5 product that FarmVibes provides matches the geometry of the product provided by Planetary Computer (in case of data_ingestion/weather/download_era5) or CDS (in case of data_ingestion/weather/download_era5_monthly), which is for the whole world. This is also similar for other data ingestion workflows (for example, Sentinel-2 or Landsat, in which the output products might match the geometry of a tile).

This is a design choice to enable caching and avoid having to redownload a product for each input geometry provided by the user. In cases, you submit a new workflow with a slightly different geometry than a previous run, we can leverage what was downloaded before.

In your case, you can clip the raster to your input geometry when reading it:

import xarray as xr
import geopandas as gpd

# Using only the first product in the time range
product = wf_run.output["downloaded_product"][0]

# Read with xarray
ds = xr.open_dataset(product.assets[0].path_or_url)

# Set CRS in case it is not defined
ds.rio.write_crs("EPSG:4326", inplace=True)

# Clip to your geometry
gdf = gpd.GeoDataFrame(geometry=[geometry])
clipped_ds = ds.rio.clip(gdf.geometry)

Let me know if you have any more doubts.

[rafaspadilha]

Closing this discussion for now. Feel free to reopen it, @ReisQuarteu, if you have any more doubts.