I am a master student and am using the SDR model for my project. To better understand my results, I am following the “Introduction to the Natural Capital Project Approach” course with the sample data from Gura. The goal is to have a base line scenario. However, both with the Gura sample data and my own data from southern Quebec, the output raster layers (like usle and export) are defined in streams and water bodies, are clear (no pixels) in most parts and have a variety of values concentrated in one portion of the territory. For example, the sed_export_Gura layer (and with my data it does the same), the streams have a value of 0 (which makes sense), while most of the territory besides a top portion has no value (clear). The top portion of the territory seems like it was modelled successfully but the rest only contains scattered pixels of random values besides the streams, which all have a value of 0. I am wondering why the portions of the territory other than the streams are not showing properly (see screenshot) like the top portion.
I tested displaying the raster with both render types palettes/unique values and with single band pseudocolor but in both cases only the top portions are correct. I am using QGIS (3.16 LTR) and InVEST (3.9) on Mac. All seemed to have gone well while running the model but I have attached my log for reference. What sort of problem might this be? InVEST-Sediment-Delivery-Ratio-Model-(SDR)-log-2022-02-01–15_09_49.txt (36.5 KB)
Hello! Thank you for your quick answering. I just tested my data with the new invest version and still have the same problem. With a render type of paletted/unique values to view the raster, only the streams and drainage show up, and only a small portion on the top of the territory shows diverse pixels, while the rest of the territory is blank. When viewed as a singleband, all is black. I will test the Gura data but I am guessing I should have had a different result with my data. I am still unsure of the nature of my problem.
Okay thanks for testing that out. I’m not sure where the problem lies either. How does the data in intermediate_outputs look?
Since these data are floating point (continuous values) paletted/unique might not be able to handle all the unique values, so it’s probably not a great choice for visualizing. Singleband is a good choice, but depending on the distribution of values, it might be hard to see if any pixels are in fact not black. What are the min/max values listed when you choose singleband? If you change them do you see more variation in the image? There could be a single very high/low pixel value skewing the colormap.
If you’re still stuck feel free to share your input data (google drive, dropbox, etc) and we can run the model on our end.
This was my problem! I reran the Gura sample this morning and changed the min-max values and got the same results as in the course!! Then I did the same for my data and I can now view my results clearly! Thank you so much!!