Apologies for what might be a simple problem for some, I’m transitioning from downloading ERA5 reanalysis as netcdfs to grib files. When I download mean surface net short wave radiation flux at hourly resolution using an API request and .grib file, the values are significantly different than those I downloaded at 6-hour resolution in August of 2024 using the online form and netcdf. This occurs particularly in the summer when flux values can be 10x as high in the hourly data vs the old 6hr data. All other variables I downloaded (mean sensible heat flux, 2m temperature, mean precipitation rate, wind speeds, etc) are pretty close but with expected changes due to higher temporal resolution. Was there a change in how this product was produced, or might I be missing some other change?