From EAB:
swfscDAS is retaining very short segments, e.g., recorded as 0 - 0.1 km that have no sightings on them, thus creating multiple extra segments that further inflate our already zero-inflated database. We have been removing these segments post processing but we were hoping you could include an option to eliminate them from the segdata output file. In our old segchopr code, we changed all zero segment distances to = 0.1 km (we need non-zero distances to compute our offset term) and delete these short segments if they do not include a sighting. It would be great if swfscDAS provided this option.
RE the short segment issue, attached are three files to illustrate the problem. The example is only for one day, but you will get the picture. On lines 115-122 of the DAS file there are multiple R to E events that resulted in three segments with 0 distance and one segment with dist = 0.0725km (see the resulting current segdata file; I highlighted the short segments in yellow). These four segments are all < 0.1 km and none of them include sightings. Our old segchopr program checked for these short segments and deleted them if there were no sightings (see "What we Want" example). The only caveat for which I did not provide an example - let's say one of these four segments DID include a sighting. We would retain the segment with the sighting but set the segment length to 0.1km. Hope this makes sense!
From EAB:
swfscDAS is retaining very short segments, e.g., recorded as 0 - 0.1 km that have no sightings on them, thus creating multiple extra segments that further inflate our already zero-inflated database. We have been removing these segments post processing but we were hoping you could include an option to eliminate them from the segdata output file. In our old segchopr code, we changed all zero segment distances to = 0.1 km (we need non-zero distances to compute our offset term) and delete these short segments if they do not include a sighting. It would be great if swfscDAS provided this option.
RE the short segment issue, attached are three files to illustrate the problem. The example is only for one day, but you will get the picture. On lines 115-122 of the DAS file there are multiple R to E events that resulted in three segments with 0 distance and one segment with dist = 0.0725km (see the resulting current segdata file; I highlighted the short segments in yellow). These four segments are all < 0.1 km and none of them include sightings. Our old segchopr program checked for these short segments and deleted them if there were no sightings (see "What we Want" example). The only caveat for which I did not provide an example - let's say one of these four segments DID include a sighting. We would retain the segment with the sighting but set the segment length to 0.1km. Hope this makes sense!