Add lightning flash extent density (FED) verification

[natalieb-noaa]

• For now, apply to HRRR / conus domain. Will want to run it on RRFS too (and eventually expand domain).
• Compute FED from HRRR column integrated graupel mass* (see below)
• Use GOES GLM data as obs
• interpolate to same resolutions as reflectivity
• use thresholds of >0, 1/4, 1/2, 1, 2, 4, 8 flashes/min (5 minute average at end of hour)

*Algo for FED: Should be able to use COLG field from model. Per Amanda: multiply that field by 3000*3000 to get kg in each model column, compute sums over 5x5 grids, and then compute flash extent density= a+b*tanh(c*(graupel mass)-d), where b, c, and d are given in the attached poster--note different values for HRRR and RRFS--and a=b*tanh(d) so that the function passes through (0,0)--on the poster David gave approximate values of (a,b,c,d), but approximating to a few decimal places like that in our code will cause 0 graupel not to map exactly to 0 flashes as it must.

AGU Conference Poster 48 36.pdf

[natalieb-noaa]

pieces of this done. currently working on the GLM data interpolation script.

[natalieb-noaa]

GLM obs interpolation part done.

update from DSG: they are doing some refactoring of the code to generate the GLM data product, so it is not ready to run in production yet.

[natalieb-noaa]

also have scripts working for model data processing and deriving FED from TCOLG. currently working on the verif script.