With the uploaded Practice info file we defined the period of harvest through H_START and H_END. Most of the Harvest dates, detected from the L4C processor, are within this period. But some (in our case 425 of 86.854 Harvest detections in our test area) are not in this period. In which period will harvest detection be conducted and how is this period defined?
Normally, in the output csv files is saved the the first harvest detection after H_START. Single detections for a parcel that are performed before the start if the week containing H_START are marked as NR in the output file.
The situations when the harvest detection is not within H_START and H_END might be:
- Harvest detection is smaller than H_START but it is in the same week with H_START (not possible to have harvest detections less than the start of the week containing H_START date => these will be marked as NR)
- Harvest detections with date greater than H_END as all the provided input times series are considered (even if the acquisitions are after H_END). This means that the first harvest for that parcel was identified after H_END (we do not have a constraint on that in this moment). This might happen from 2 reasons:
- The harvest was really performed at that moment (I assume is less likely)
- there is an issue with the input data - most likely gaps in the S1 pre-processed time series. In this case you should check the S1 but also the optical data to see if there is everything OK in the pre-processing chain (a first check would be in the “Statistics” tab of the system website).