ARCSI v4.0.0. Missing data

[Pauline Burke]

Hi Pete,

Our testing of ARCSI v4.0.0 has shown that in one case, the northern 35km of data has been 'chopped' off a scene in T30VNX covering the north of the Shetland Islands. In the same input granule run through v3.1.7 this portion is not missing. The cut-off line seems to align with latitude 61N.

Secondly, we have noticed some differences in the topographic shadow masks from the two versions in all test data. We are running S2CLOUDLESS for cloud masking in v4, whereas v3.1.7 is obviously using Python Fmask - would this have an influence? 

Thanks,

Pauline

[Peter Bunting]

Hi Pauline, 

I am sorry I haven’t really had chance to look at your previous issue - I’ve got project deadlines I am behind with.

Not sure why part of a scene would get removed. You could get things like that if the DEM had no data regions but as it runs through with the older version I am not sure of the top if my head. Can you send me the scene where this is occurring and the DEM you are using?

I don’t think the topographic shadow mask should have changed. What type of differences are you seeing? Does it change without using the cloud masking?

Best wishes, 

Pete

--
You received this message because you are subscribed to the Google Groups "RSGISLib Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rsgislib-suppo...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/rsgislib-support/5407c2e7-5a50-49d9-836d-6d4618b8a54cn%40googlegroups.com.

[Pauline Burke]

Hi Pete,

Just a bit of feedback on the above. I tested ARCSI v3.1.7 and v4.0.0 without cloud shadow masking (so essentially running with completely identical parameters), and the differences in the topographic shadow mask still exist and are the same in location and magnitude. Perhaps this then points to differences in the post-filtering you mentioned last week.

Best Wishes,

Pauline