I’m seeing small differences between ECMWF Open Data and MARS. For example, when I compare the temperature data at p = 1000 hPa for step=24h at 0.25 degrees resolution between ECMWF Open Data and MARS, there is a tiny difference like below. The difference seems a noise rather than a systematic difference. I’d like to confirm whether this is expected (e.g., due to grid/interpolation/packing) or if I’m comparing the products incorrectly. Note that ECMWF Open Data is given with 0.25 degrees resolution while MARS data is regridded to 0.25 degrees when downloading.
