Tme series data don't overlap after filtering in TMS

[pritamd47]

TMS can receive time-series that overlap in time when passed into the function, but may get unaligned after filtering out the suspected erroneous points. This case is not handled in TMS currently.

RAT/src/rat/core/sarea/TMS.py

Lines 359 to 368 in a85f1d6

	area['filtered_area'] = deviation_from_sar(area['area'], sar['area'], AREA_DEVIATION_THRESHOLD)
	area.rename({'area': 'unfiltered_area'}, axis=1, inplace=True)
	# area.rename({'filtered_area': 'area'}, axis=1, inplace=True)
	area_filtered = area.dropna(subset=['filtered_area'])
	area_filtered.loc[:, 'days_passed'] = area_filtered.index.to_series().diff().dt.days
	area_filtered.loc[:, 'trend'] = area_filtered['filtered_area'].diff()/area_filtered['days_passed']
	sar, area_filtered = clip_ts(sar, area_filtered, which='left')

Example:
Using default Area-Deviation threshold (5%), for the Taylor park reservoir (Gunnison, CO) in the test suite, the threshold is very low (0.37 sq. km.). This causes most of the areas to get filtered out and the resultant dataframe's max date is less than the SAR's min date. This results in an empty dataframe when clip_ts is used. Currently, the hotfix was resolved by increasing the AREA_DEVIATION_THRESHOLD value for smaller reservoirs thru 03a1b8c. The new default thresholds are as follows:

• 5% for reservoirs with area >= 100 sq. km.
• 10% if 100 > area >= 10 sq. km.
• 25% if area < 10 sq. km.