slice cutting shortcoming

[renierlgv]

I've noticed that due to differences in S1 slice cutting, an area including N bursts can have one burst with only a handful of dates. T his poses a problem for long time series because longer time spans allow for more permissive amplitude dispersion threshold, meaning that for the burst with the few dates of data a significant part of the points will be wrongly considered PS. What is recommended here? should the handful of dates be removed from all bursts (not critical only in the case they are at the beginning of the time series) or the issue of the artificially low AD due to the small amount of data will be suppress when combining with the rest of the data?

[scottstanie]

the OPERA approach was to use the following function:
https://github.com/opera-adt/opera-utils/blob/main/src/opera_utils/missing_data.py#L54

which is in the download module here for the geocoded CSLC product using the check_missing_data flag:
https://github.com/opera-adt/opera-utils/blob/main/src/opera_utils/download.py#L180-L215

it removes dates that are outlier "missing data dates", or (for very large areas) would exclude a spatial Burst ID that is missing from most of the dates