Runoff variable from land dataset near coasts of ERA-Interim; land-sea fraction applied or not?


When I pull lat-lon grid runoff for ERA-Interim, how is runoff handled in your model near coasts?  I understand that fluxes (latent heat flux, e.g.) might be a fraction of what was computed on native grid from land and sea (so, weighted sum) once the flux field is provided on lat-lon.  (Correct?)  But, how is runoff handled?  Say I have ERA-Interim runoff for a river near a coast, and the lat-lon data show all water for that river reach, with runoff=-9E+33; is there another dataset (i.e., perhaps on a different grid like one of the Gaussian grids or something?) that might provide more information with respect to any runoff that might have been computed for that river reach/basin, but might have been discarded, somehow, in reprojecting fields (like runoff) to the lat-lon grid?  Or, is the output always just the depth for grids that have majority land (so, binary "1" for output land-sea mask) and -9E33 for minority land grid cells, and all bets are lost for river reaches near coasts?

What are the answers to these questions with respect to the ERA5 land runoff data?

Thank you in advance,


Does anyone have an answer for this?

If it helps, in other words:

Simply, what is the volume of water for the pixel?  Depth x area of pixel x land fraction, or depth x area of pixel (which would assume that some average depth of runoff for the area was already computed on your end)?

Is runoff of any pixel with any land fraction included?  Is this the case for any product/grid?  (Or, are some products'/grids' files using a threshold of 50% land, e.g.?)

Meanwhile, is seems some products use a large negative value to represent water pixels; for which is this the case?