search

CH-Earth / summa

Structure for Unifying Multiple Modeling Alternatives:
http://www.ral.ucar.edu/projects/summa
GNU General Public License v3.0
80 stars 106 forks source link

scalarTotalET sometimes not being calculated correctly #327

Closed arbennett closed 2 years ago

arbennett commented 6 years ago

While making some water balance plotting scripts I came across some inconsistency in how the variable scalarTotalET is output . Some HRUs calculate this value correctly while others just have zeros.

The correct calculation (per the SUMMA code) should be: scalarTotalET = scalarGroundEvaporation + scalarCanopyEvaporation + scalarCanopyTranspiration

Below is a plot showing one of the locations where this issue was present. You can see that the manual sum of the components is simply the ground evaporation in this case, but the derived scalarTotalET_mean is zero across the timeseries.

totalet_issue

This does not appear to be soil or vegetation type related as far as I can tell. Can anybody else reproduce this issue?

bartnijssen commented 6 years ago

@arbennett: Are these SUMMA calculated means? Can you reproduce this with the instantaneous values? Just so we can isolate the problem a bit more to either the real SUMMA calculation or an error in the calculation of the daily means.

arbennett commented 6 years ago

Those were SUMMA calculated means. It appears reproducible with instantaneous values. Below is a plot.

totalet_issue_timestep

bartnijssen commented 6 years ago

@arbennett: Given that this does not happen for every cell in your Columbia domain - can you include any additional info that makes this grid cell different? (veg type, soil type, always snow covered, never snow covered, etc.). Also - would it be possible to update the plots with actual dates rather than day numbers? Not clear whether day number 200-300 is winter (and hence snow covered?)

arbennett commented 6 years ago

I haven't been able to identify anything unique about these locations. They don't share a common vegetation type, and while they all share the same soil type (clay), there are HRU in this run that have those characteristics and calculate the scalarTotalET as expected. None of the issue sites are either covered nor empty of snow all the time. There are 9 of 24 in the range 7009-7032 which show this issue. They are [7010 7012 7013 7015 7017 7018 7021 7028 7032].

I should also note that this is reproducible for me when running a single problem HRU, instead of running the entire 24 HRU batch.