DCAPE Values about sharppy HOT 5 CLOSED

One key difference that connects to the DCAPE calculation is the selection of the lowest Theta-E value in the lowest 400 mb of the sounding. Comparisons between the SPC sounding page images and SHARPpy shows that sometimes SHARPpy selects a completely different level as opposed to SPC. This is especially seen in the example above. Because the parcel origin is significantly lower in SHARPpy than in the SPC image, the DCAPE will be lower than SPC's.

Step 1 to solving this problem is fixing the correct selection of the parcel. Without correcting that (if it needs correcting) attempting to correct any problems with the parcel lowering portion of the code isn't going to be possible.

from sharppy.

Rich Thompson provided us with John Hart's DCAPE routine from NSHARP. The routine uses a layer average minimum theta-e to determine the downdraft parcel's point of origin. The DCAPE routine also does not include any virtual temperature correction, so this will have to be changed.

from sharppy.

I implemented the DCAPE code from NSHARP into SHARPpy, and the DCAPE calculations compare significantly better to the SPC DCAPE indices. There still are differences that arise though between the calculations, however they are on the order of +/- 10 J/kg, with the most that I've seen around 50 J/kg. Other times the calculation is spot on. I still believe this is a consequence of us not picking the correct downdraft parcel, however I have a suspicion that this is not a consequence of the code, but rather the averaging method. Right now I am using pressure-weighted averaging to compute the mean theta-e values, however NSHARP may be using the exact averaging instead. This will need to be investigated further, however the pressure-weighting is more accurate, so changes may not be necessary.

from sharppy.

I tested it with a couple of soundings, and like you said: some are spot on, others slightly off. I'm going to close this issue since we're in the right range of values. I'll likely start looking for some places to optimize later on.

from sharppy.

The sounding generation process is taking close to 3 seconds now, up from 1. So some profiling of the function will be needed in the future.

from sharppy.