Help with SDR and sdr max values

I am in the process of running the SDR model for a big watershed and I ran it once and got this. error code after the model completed: watershed.shp does not intersect with the raster (u’D:\sdr_workspace\sed_retention.tif’, 1)
2019-10-01 22:20:44,276 model._logged_target(1642) INFO Execution finished

I am running it for a fairly large watershed. I will attached the log too.
Also I am wondering what an appropriate SDR value is? Not entirely sure what it is supposed to represent as I can’t find any documentation on it. InVEST-Sediment-Delivery-Ratio-Model-(SDR)-log-2019-10-01–16_23_44.txt (236.1 KB)

Hi @sfix -

Thanks for posting your log file. It must be a big watershed, to take almost 6 hours, and I’m surprised that it took so long to give this error. It seems to be happening when the model is summing the raster results by your watershed shapefile. Given the error message, my first guess is that bigSioux.shp has a different coordinate system than the other inputs, so check to make sure that it has exactly the same projected coordinate system as your raster inputs.

Another possibility is that some of the watershed polygon (or some of the sub-watersheds, if you’re using those) lies outside of the raster coverage, so the resulting rasters do not cover the whole watershed. In this case, you’ll need to either extend the input rasters to cover the whole watershed, or choose a sub-watershed that lies within the raster coverage.

It does appear that the model actually produced all of the raster results. So if you’re only interested in those, you might not need to worry about the aggregation error. Another nice thing is that the new feature Taskgraph might make re-running the model go much faster, since it won’t re-run any of the calculations that haven’t changed since the previous run.

By “SDR value” do you mean the “Max SDR Value” model input? The User Guide says this:

SDRmax (required): The maximum SDR that a pixel can reach, which is a function of the soil texture. More specifically, it is defined as the fraction of topsoil particles finer than coarse sand (1000 μm; Vigiak et al. 2012). This parameter can be used for calibration in advanced studies. Its default value is 0.8.

Does that help? I’ll say that I’ve run this model a lot, and have always used the default value of 0.8.

~ Stacie

Hi Stacie,

I did check the projections, and one was wrong. Then I fixed that one so now they are all on utm 15n. But it is still not working. All of the rasters exceed the watershed shapefile boundary. sdrtest3.txt (3.6 KB)

Hi @sfix -

Well I’m not sure what’s happening, so if you can send me your data I’ll check it out - swolny@stanford.edu.

~ Stacie

Hi Stacie,

Sent you an email.

Thanks!

Hi @sfix -

Looking at your data in ArcGIS and QGIS, and I see the same thing - the shapefile bigSioux.shp does not overlap the rest of your input layers, it’s hanging out alone to the east of the other layers. Have you tried running SDR with BSwatershed.shp instead, which does overlap the raster layers? Do you get the same error?

I’m not sure why they don’t overlap, since the coordinate system seems to be the same. How did you set the coordinate system? Did you do a Reproject/Warp, or some other process?

~ Stacie

Hi Stacie,
I set the coordinate system by re-projection. Not sure why it doesn’t work, it looks like it all overlaps on my ArcMap. I tried running the model with the BSwatershed.shp, and it seemed to work, but then half way through, it crashed. Any tips on getting it to run? Should I reduce the quality of the k-factor input? Find a better computer?

Glad it got further with the other watershed. Did the whole SDR app crash? Or did you get an error of any sort? Does the log file give any indication of what happened? Were any outputs produced?

For SDR, all inputs are resampled to be the same resolution as the DEM, so it’s the DEM that really matters for setting the size of the files that are processed. InVEST is rather good at memory use, and handling large files, although of course the hardware/OS can be a limiting factor. It looks like your DEM is 2.7GB, which means that all of your data outputs will be that large, and that can really stress a computer.

~ Stacie

Hi Stacie,

The whole app crashed. I am trying to run it again on a server so I will get back to you on the the log if there is one.

Hi Stacie,

My apologies for the delayed response. Here is the txt file from the crashed run.InVEST-Sediment-Delivery-Ratio-Model-(SDR)-log-2019-10-03–13_30_23.txt (44.4 KB)