Freezing Level Calculation in UPP

[WenMeng-NOAA]

A GFS user has an inquiry about freezing level calculation for GFS/GEFS products as follows:

It appears the GFS has issues with their freezing level computations. We've noticed odd oscillations when there are cold periods. I am not sure if the native GFS freezing level procedure is "ground up" method (I think it is, but not 100%). However, we do a "top down" method with the appropriate fields in our procedures, and it still is an issue for some basins depending on the situation. It is mainHey Brettly in the Sierra. The issue seems to be more pronounced with the native GFS freezing level estimates, but it still is an issue with our "top down" method. The ECMWF doesn't appear to have this issue from what we've seen in operations. I believe the native freezing level field in the ECMWF is "top down". Talking to others in our office, we suspect it has to do with extrapolation below the ground surface. We've tried to account for this in our operational procedures (made the change last week). The odd behavior seems to be mostly on clear days, but I am not 100% it is not an issue for all precip situations. Since MEFP lumps everything together, the parameterization could be dominated by the diurnal behavior, and applied to all situations. For the third parameter enhancement, I suppose we could screen the data for precip only conditions (positive positive), but I think the ideal solution is to correct the issue in the GFS. This was just discovered recently, so we need to do more research. I'll touch upon this in my presentation on Wednesday. Below is one example from CW3E that shows freezing level from the GFS compared to the ECMWF for the Upper San Joaquin. You can view other watersheds from their page: https://cw3e.ucsd.edu/DSMaps/DS_freezing.html.