Introduction

The Copernicus European Regional ReAnalysis Land (CERRA-Land) provides spatially and temporally consistent historical reconstructions of surface and soil variables at the same horizontal resolution as the CERRA high-resolution reanalysis from 1984 to the present.

The CERRA-Land Analysis Ready Cloud Optimised (ARCO) data presented are a subset of surface and soil variables from the CERRA-Land sub-daily regional reanalysis on a regular latitude-longitude grid at a 0.05ºx0.05º resolution and are stored in a ARCO format. It allows direct access to the single levels variables (see below) without downloading individual files, enabling efficient and scalable data access and retrieval. The ARCO format has a dual-chunking approach to optimize data retrieval for long periods of time over small areas (geo-chunked) or short periods of time over large areas (time-chunked). 

Methodology

The process to generate this dataset involves:

• Fetching CERRA-Land data in GRIB format.
• Regridding from the native Lambert conformal conic coordinate system to a 0.05ºx0.05º regular latitude-longitude grid using conservative remapping. 
• Applying homogenization conventions, as applicable:
  
  • Ensure latitude, longitude are stored in a growing fashion. Ensure that longitude format has negative and positive values, instead of only positive. The dataset has extent:

    • longitude: [-60, 75]

    • latitude: [15, 80]

• Keep only 3-hourly data. This does not apply to total precipitation which is a 24h total accumulated parameter.
• Writing to an ARCO Zarr archive (which is similar in structure to netCDF).
• The CDS requests the data from the Zarr archive using xarray and writes the data to netCDF or CSV (as requested).

Data organization and access

The ARCO data is stored as Zarr datacubes to provide efficient access to time-chunked and geo-chunked ERA5 pressure levels 6-hourly data:

The Zarr datacubes can now be accessed using your CDS API key. Below is a simple plug-and-play code snippet and a more detailed Jupyter Notebook example. For more detailed and advanced access examples, please refer to the main Analysis Ready Cloud Optimised (ARCO) Data.

import xarray as xr

# Geo-chunked data for access optimised along the time dimension (e.g. for time-series at a single point)
# Surface
geo_url = "https://arco.datastores.ecmwf.int/cadl-arco-geo-045/arco/reanalysis_cerra_land/surface/geoChunked.zarr"

# Time-chunked data for access optimised along the time dimension (e.g. for short-time period for large areas)
# Surface
time_url = "https://arco.datastores.ecmwf.int/cadl-arco-time-045/arco/reanalysis_cerra_land/surface/timeChunked.zarr"

# Open one of the Zarr objects with xarray, the default example opens the geo-chunked surface variables
ds = xr.open_zarr(
    surface_geo_url,
    consolidated=True,
    storage_options={
        "headers": {"Authorization": f"Bearer <CDS-API-KEY>"}
    }
)

# Inspect the variables
print(ds)

Time-series access via the CDS

The CERRA-Land time-series data on single levels from 1984 to present catalogue entry allows users to download long time series for a single-point or limited area via the CDS download webform or a CDS API request that users can build from the webform. The data download via the time-series entry is as described below with additional aspects:

1. Please be aware when you select a "Location" on the form, the latitude and longitude values are rounded to the closest neighborhood point of the 0.05 degrees grid. Similarly, selecting an “Area” returns all points within the region defined by the nearest neighborhood corner points to the user-specified S, W, N, and E latitude/longitude values (maximum 5° x 5°).

2. The download form allows users to choose netCDF or CSV. The conversion is done directly with xarray with no additional processing from the Zarr description above.
   • Given the computational cost of generating CSV files, users are limited to downloading a small volume of data with this option.

3. Please note that a request with a forecast parameter and an analysis parameter will return two separate files.

Spatial grid

The parameters are on a regular horizontal grid of 0.05º.

Temporal frequency

Temporal frequency of the data is 3-hourly for surface and soil parameters and daily for total precipitation.

Data update frequency

Data will be added half-yearly (in June and December)

List of Parameters

Table 1 shows the surface and soil level parameters for which time-series data can be retrieved. From the available parameters in the CERRA-Land dataset, only 5 of them have been chosen to be included in this time-series dataset. Some others might be added depending on the user requirements. Apart from the total precipitation which is a variable of type analysis, the others are of type forecast.

Table 1: List of available parameters

evaporation

surface_latent_heat_flux

surface_sensible_heat_flux

total_precipitation

volumetric_soil_moisture

Known issues

Please refer to the CERRA-Land documentation.

Related articles