Page 1 of 1

Aggregating runoff data in a hydrological catchment

Posted: Thu Mar 05, 2020 2:51 pm America/New_York
by GES DISC - jimacker
The GES DISC Help Desk recently received this question from a user, and it was answered by a Subject Matter Expert. It is provided here for information content.

I have a question regarding the runoff data from the GLDAS dataset and the way
I should aggregate it within a hydrologic catchment.

I have downloaded the mean monthly runoff data (Qs_acc & Qsb_acc variables from GLDAS Noah Land Surface Model L4 monthly 0.25 x 0.25 degree V2.0) and one of the downloaded netCDF files is attached for your consideration. I want to aggregate the total runoff data in a catchment and compare it with observed runoff values from a hydrologic station located in the downstream of the catchment. Therefore, I have first got spatial subsets of storm runoff (Qs_acc) and baseflow (Qsb_acc) grids within the catchment and then summed them up to get the total runoff values. Then I calculated monthly average values for the whole area (by averaging total runoff from all existing grids in the catchment for each month). This way, I guess that I'll get a time series of monthly runoff estimations in 'kg m-2' which is equal to 'mm'.

However, I get very low runoff values (between 0-1 mm/month). The Vmax values in the attribute table of the netCDF files for Qs_acc and Qsb_acc are given as 5.6851 and 3.7703, respectively while the corresponding observed runoff values are around 10-50 mm/month. Do you have any idea about the possible underestimation of runoff values in GLDAS dataset or maybe my calculation is wrong?

Any tips or advice in this regard is much appreciated.

Re: Aggregating runoff data in a hydrological catchment

Posted: Thu Mar 05, 2020 2:52 pm America/New_York
by GES DISC - jimacker
Here is the answer, a collaboration between our SME and a GES DISC staff scientist.

1. GLDAS total runoff is the sum of subsurface runoff “Qsb_acc” and surface runoff “Qs_acc”.

2. Since the units of “kg m-2” as 3-hour accumulation is correct, there is no near term plan to change the units in the NetCDF, having "per 3-hour" in Readme should help. However, she plans to change the “cell_methods” to "cell_method: time sum" for the *acc fields in the 3-hourly files in future reprocessing. For monthly data, "cell_method: time mean" is correct as it is average of 3-hour files. Thanks a lot for your suggestions.

3. Regarding the snowmelt, its contribution in the runoff fields are already accounted for, so it doesn't need to be added to total runoff. It's provided as a separate diagnostic field for snowmelt by itself.