Closed EsmailGhaemi closed 7 months ago
Attention: 1 lines
in your changes are missing coverage. Please review.
Comparison is base (
c2f8320
) 83.34% compared to head (a2b6761
) 83.34%.
Files | Patch % | Lines |
---|---|---|
pysteps/verification/salscores.py | 94.73% | 1 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Thanks @EsmailGhaemi for the changes. Could you just spend few more words on what you did? It's not obvious to me what you changed, why, and what kind of impact we should expect. This will help to describe the changes in the release notes.
Yes, of course. In the previous version, the scaled volume was calculated for each object separately, and then in sal_srtucture function we summed all of them. However, when calculating total volume (V_R in equation 8, Wernli et al., 2008), we should multiply each object's scaled volume (v_n) to the amount of precipitation for that object (R_n) as well. Now the scaled volume function directly returns total scaled volume (instead of scaled voulme for each object). Additionally, for the weighted distance function, np.nansum() is used to avoid the effect of nan values.
Brilliant, thanks, I've update the description above.
Fixed #344 (S parameter edited, L parameter modified for dealing with nan values)
Now the scaled volume function directly returns total scaled volume (instead of scaled voulme for each object). Additionally, for the weighted distance function, np.nansum() is used to avoid the effect of nan values.