ERA5 data on the Planetary Computer¶

The ERA5 dataset from the ECMWF, converted to Zarr by PlanetOS data on weather.

Query using the STAC API¶

The data assets are a collection of Zarr groups, which can be opened with libraries like xarray.

Each STAC item covers a single month and the entire globe. You'll likely query on the datetime.

In [1]:

import pystac_client

catalog = pystac_client.Client.open(
    "https://planetarycomputer.microsoft.com/api/stac/v1/"
)
search = catalog.search(collections=["era5-pds"], datetime="1980-01")
items = search.get_all_items()

len(items)

Out[1]:

In [2]:

list(items)

Out[2]:

[<Item id=era5-pds-1980-01-fc>, <Item id=era5-pds-1980-01-an>]

There will be two items per month, depending on the kind of variables within: fc (or "forecast") and an (or "analysis").

An analysis, of the atmospheric conditions, is a blend of observations with a previous forecast. An analysis can only provide instantaneous parameters (parameters valid at a specific time, e.g temperature at 12:00), but not accumulated parameters, mean rates or min/max parameters.
A forecast starts with an analysis at a specific time (the 'initialization time'), and a model computes the atmospheric conditions for a number of 'forecast steps', at increasing 'validity times', into the future. A forecast can provide instantaneous parameters, accumulated parameters, mean rates, and min/max parameters.

You can narrow your search to a specific kind using the era5:kind property.

In [3]:

import pystac_client

catalog = pystac_client.Client.open(
    "https://planetarycomputer.microsoft.com/api/stac/v1/"
)
search = catalog.search(
    collections=["era5-pds"], datetime="1980-01", query={"era5:kind": {"eq": "an"}}
)
items = search.get_all_items()

print(len(items))
item = items[0]

There are several assets available, one for each data variable. We can build up a dataset with all the variables using xarray.open_dataset and combine_by_coords.

In [4]:

import planetary_computer
import xarray as xr

signed_item = planetary_computer.sign(item)
datasets = [
    xr.open_dataset(asset.href, **asset.extra_fields["xarray:open_kwargs"])
    for asset in signed_item.assets.values()
]

ds = xr.combine_by_coords(datasets, join="exact")
ds

Out[4]:

<xarray.Dataset>
Dimensions:                            (time: 744, lat: 721, lon: 1440)
Coordinates:
  * lat                                (lat) float32 90.0 89.75 ... -89.75 -90.0
  * lon                                (lon) float32 0.0 0.25 ... 359.5 359.8
  * time                               (time) datetime64[ns] 1980-01-01 ... 1...
Data variables:
    air_pressure_at_mean_sea_level     (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    air_temperature_at_2_metres        (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    dew_point_temperature_at_2_metres  (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    eastward_wind_at_100_metres        (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    eastward_wind_at_10_metres         (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    northward_wind_at_100_metres       (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    northward_wind_at_10_metres        (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    sea_surface_temperature            (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
    surface_air_pressure               (time, lat, lon) float32 dask.array<chunksize=(372, 150, 150), meta=np.ndarray>
Attributes:
    institution:  ECMWF
    source:       Reanalysis
    tilte:        ERA5 forecasts

Now we can plot timeseries for the month, averaged over space:

In [5]:

ds["sea_surface_temperature"].mean(dim=["lon", "lat"]).plot();

Or make a map of some variable on a single date.

In [6]:

ds["sea_surface_temperature"].isel(time=0).plot(size=8);