Working with satellite swath data

[OscarGuzman]

Hello,

I am working with some TRMM/GPM global datasets to model the precipitation produce by hurricanes. In particular, I am using both 2A products of GPROF retrievals that originally come in HDF5 format.   By working with python coding, I was able to extract a list of Lat/Lon coordinates with the outer envelope of the satellite’s swath, with the aim to create a polygon vector that allows me to perform some basic map algebra operations over my hurricane database.

However, since each swath crosses both poles and the international dateline, the conversion from points to line (and polygon) leads to an erratic representation of the swath in those areas, as shown in the attached figures.

Any hint about how to work with global satellite swaths? At this point I just need the polygon with the envelope.

[PeterBecker]

This appears to be more of a polygon geometry question vs one related directly to remote sensing, but certainly something that needs to be resolved. Not sure why you are getting the polygons intersecting as they are. I assume you are closing of the strip, but Can you clarify how you are creating the polygon and what projection your data is in. Assuming you have first created a polyline that represents the extent of the scene then there are multiple ways to create the polygon. Due to the overlap at the end this is a self intersecting polygon. Please try out the method defined in https://support.esri.com/en/technical-article/000021197

If that does not work can you provide a link to the line feature and we can take a look at how best to convert to a suitable polygon.

[OscarGuzman]

Hello,

Thanks for your help and interest in this topic. You are right; at first glance, it seems to be more a polygon-topology related problem than a Remote Sensing issue.

Yes, as you mentioned, I am converting my list of swath latitude/longitude coordinates into a Polyline first and then into Polygon.  The list of points is arranged in a counterclockwise sequence with the same start and end coordinates to allow a perfect closing either of a line or a polygon. (i.e., like a connect the dots game)

I hypothesize the problem arises when the Point to Line conversion crosses the international dateline, a place where the longitude suddenly changes from -180 to + 180 (or vice versa). Please notice that a line appears from west to east at the latitude where the satellite swath finds the +-180 longitude.

I tried the same procedure using swaths from different satellites of the GPM constellation and changing from WGS84(EPSG 4326) to pacific-centered coordinate systems, obtaining a similar result. But each time having the west to the east line starting at the limits of the extent of each cartographic projection.

Please find a shapefile (Point) with the swath on this link:

https://wetransfer.com/downloads/7a1d38c6f2e617e869dab13a1e0f030920200826004807/514b569d725f7b7bbbfa...

I hope it helps to understand the problem.

[PeterBecker]

We had a look at this. The solution is to create it as a polygon with two parts. You need to rearrange the points into two parts and change the point orders. One part starts at approx 180,89 and goes round to approx -87,-136 and returns to 180,68 and then closes. The second part should start at -180,68 goes to around 43,72 and then return to -180,89 and close.To ensure a clean cut at the date line you may want to add vertices with coords at 180 and -180. In this way you will get a polygon with two parts, but covering the correct area.