carbonplan / cmip6-downscaling

Climate downscaling using CMIP6 data
https://docs.carbonplan.org/cmip6-downscaling
MIT License
167 stars 24 forks source link

Missing data in ERA5 #290

Open arbennett opened 1 year ago

arbennett commented 1 year ago

:wave: Hi, I'm working with @orianac on using this data for some ML-based snow projections and noticed that there is some missing data in the ERA5 data. Specifically, pr is missing for 1985 and tas is missing for 1980 and 1996. I think this is for the entire domain, but my initial searches were just in the Western North America Region. You can ignore the line for SWE, that's from my own data.

image (9)

norlandrhagen commented 1 year ago

Hey there @arbennett, thanks for reporting this. Do you mind providing a minimal reproducible example? We do have a few copies of the ERA5 data, so I could totally see us having some missing chunks.

A quick look of what I'm seeing:

import intake
from cmip6_downscaling import config
import xarray as xr

cat = intake.open_esm_datastore(config.get("data_catalog.era5_daily.json"))

ds = xr.concat(
    list(cat.search(cf_variable_name=['tasmax','tasmin','pr']).to_dataset_dict().values()),
    dim='time',
)

ds.pr.isel(lat=0,lon=0).isnull()
ds.tasmax.isel(lat=0,lon=0).isnull()
ds.tasmin.isel(lat=0,lon=0).isnull().plot()
image image image