That would certainly be nice.

However, we'd need to change the way Meteostat works to achieve that. Currently, model data is specifically stored and retrieved only for the locations where a weather station is installed.

Maybe we can set up a grid (e.g. 5 * 5 km) and provide model-based time series for each grid cell. Then, if there are no stations nearby or gaps in the series, we fall back to the closest grid cell.

Biggest challenge would certainly be the amount of data. If we want to have a reasonable resolution, we'll end up with millions of grid cells. I doubt this is feasible without bigger funding. But I agree it would improve data availability in remote areas (or those with few stations).

Just curious, what is/are the source/s for model data?