Negative values in ERA5 runoff

Hi everyone,

Does anyone know why there are negative values in the runoff? Is it an error? I saw the topic that someone found negative values in precipitation and the answer says it's caused by conversion from grib format to netcdf. Is it the same as in this case?

I also found small difference that the total runoff (variable 'ro' in ERA5) isn't exactly the same as the sum of surface and sub-surface runoff ('sro' and 'ssro' in ERA5). Is it the same error which caused by data conversion? How are these variables calculated in the data? Is the total runoff calculated after modeling the surface and sub-surface runoff? Or does the sub-surface runoff being calculated after subtracting the simulated surface runoff from the total runoff? I'm wondering which one is directly simulated in the model. I know the surface runoff is simulated, which is written in the physical process document. How about total runoff and sub-surface runoff?

Thanks in advance,