Unable to mosaic over the dateline

Using gpt (or the Mosaic GUI) I have been unable to project/mosaic images over the dateline. I have tried several different kind of satellite images (MODISA, MERIS, …) without luck. I originally used SeaDAS 7 (see this thread on their forum http://oceancolor.gsfc.nasa.gov/forum/oceancolor/topic_show.pl?tid=6394 ), but now tried it with the newest version of SNAP as well with the same result. The output does not look at all what I expect (using stereographic projection it wraps around the pole, see image below). If I set my eastermost limit to 179 it works fine…

I use it as follows:
/Applications/snap/bin/gpt -e /Users/*****/Downloads/snap/mosaic_ross_sea.xml -Ssource=/Users/*****/Documents/snap/MER_RR__2P_TEST.N1 -t /Users/user/Documents/snap/MER_RR__2P_TEST4.N1_nc -f netCDF4-CF

Graph file attached as well. Any help appreciated…

mosaic_ross_sea.xml (1.5 KB)

The dateline makes always trouble when doing reprojection. The pole is even worse. And you have both in your data…
However, I replicated your use case and I didn’t find the result so bad.
Yes, the world view draws a scrambled shape. That’s true. I get the same result. But when you switch to the World Map Tool View it looks better. Not good but better.

What’s wrong here, is that it can’t correctly decide what is inside and what is outside of the product.
You can also see (the white shape) the input product I’ve used.
When looking at the actual data of the result it looks much better.


On the right side I see the input product at the right place. The both left images show the latitude and longitude values for each pixel. Looks also correct.
Also when moving the mouse cursor over the image I get correct coordinates.
Is it only the shape on the world map you war not satisfied with?

Thanks for helping. Indeed, the shape of the output is not what I expect. I expected the output to be limited by my latitude/longitude bounds (westBound 162.0, northBound-72.0, eastBound -156.0, southBound -79.0). It projects way to far south and east… See below image when projecting it in IDL/SeaDAS6. The output image is bounded by the lat/lon limits I specify…

OK, now I see.
I consider this as bug.
At a workaround, you can add a subset operator to your graph. The size of the output of the mosaic is known and so you can crop to a region by specifying pixel-coordinates.
I’ve attached an example graph to do this.
subset_mosaic_ross_sea.xml (1.8 KB)

Thanks for creating a bug report for it.