Overwriting Correct Speeds

[SorooshMani-NOAA]

The change at

StormEvents/stormevents/nhc/track.py

Line 1183 in fd0da54

advisory_data["speed"] = speeds * 1.9438 # m/s to kt

which was made as a part of 04bc119 results in overwriting speed values. This results in incorrect values when writing out the tracks.

It was noticed in one of the ensembleperturbation tests: noaa-ocs-modeling/EnsemblePerturbation#185

[SorooshMani-NOAA]

I noticed that in #117 we updated the speeds for the tests in this repo. @WPringle was there any reason to change the reference files? Thinking about it now, I believe we should make sure on write we just convert values back to the unit used in ATCF, right?

[SorooshMani-NOAA]

The fix seems to be just to convert inside the PC file methods rather than in the dataframe itself. I can implement this if that sounds right.

[WPringle]

@SorooshMani-NOAA Yes essentially if we just want to write out speeds in original ATCF file then they would already be knots while if we calculated internally it needed to be converted, so yes the fix should be to make sure the output from the internal calculation is just in knots.

[SorooshMani-NOAA]

@WPringle trying to change the logic, I realized the code is a bit twisted! In the initial reading of a .22 file it actually tries to recalculate the speeds anyways (because of an internal non-existing location hash) but then the calculated value is different from what is read from the input track! I guess the input track might be incorrect because of the missing speed conversion logic at the time of saving the test input 2 years ago! Given that I think I should update my "input" for the failing tests, not the logic or the output!

[WPringle]

@SorooshMani-NOAA Sounds good, as long as it always recalculate the speeds then should be fine.