As far as I understood, the L4B processor extracts its own independent marker database each time it is run. Is there any way to keep this database to see what data the L4B results are based on? I am specifically interested in the values that have been used to mark a mowing event, or the individual time series for parcels without any detected mowing events.
I aim to use this data for validation purposes since this year we are doing in-house L4B validation, and afterward want to see if there is a correlation of false negative/false positive cases and in case a correlation is established, adjust the thresholds accordingly.
This brings me to my second question - if we have established a certain correlation and wish to adjust the behavior of the processor, which parameters do we adjust in the config? For NDVI parameters I see that there are thresholds in the config. For S1, I see pfa - “false alarm parameter”. So if I am getting an increased amount of false alarms, should this parameter be increased or decreased, and if so, by what increment?
I am aware these are pretty specific questions, but I would really appreciate having some insight as to how we can adjust the l4b processor according to our field validation results.