I am running the SWY model, it works fine but I have some problems. On the one hand, the QF values show values of 0, both the annual accumulated and the qf values per month, this occurs in areas of mountain peaks.
I have noticed that the curve number maps also have gaps in the pixels where this occurs. To my surprise, this does not correspond to a single land use, but to two. But these gaps do not mean that there are no curve number values assigned to that LULC
On the other hand, I see that the values of L, L_sum, B_sum have negative values that are very large therefore relevant. I have read some things on the forum but I have not found the solution for my problem. The layers I have of inputs do not seem to have any data values. What can be caused this?
Additionally, the spatial pattern of the negative values is somewhat strange.
If details are needed I can attach a folder with the data. From already thank you very much
You also see the same pattern as QF and CN
in a layer called Yes (in the cache_dir folder), could someone explain to me what this output is about? Thank you!
Thanks for posting and providing that information. This might be easiest to review if we could reproduce it ourselves using your data. Is it possible to share your data using Google Drive or another cloud sharing service? Here’s a post about how to collect all your InVEST inputs that we’d need to run the model.
Oh, could you also mention which version of InVEST are you using?
Thank you very much for the response, the version of INVEST that I am using is 3.12.1
These are the model data, thanks for the guide.
[invest_seasonal_water_yield_datastack.tgz - Google Drive](https://SWY model )
Very grateful in advance for any help.
Would you mind downloading and running the latest version of InVEST, 3.14.0? We have made some updates and fixes, specifically to SWY that might have an impact on what you are experiencing.
After trying that please report back here and I can continue to look at your data if necessary.
Ok. Right now I’m going to download the new version and run the models. After this I comment on the results. Thanks for the help!
I have run the model with the new version of INVEST 3.14.0 but the problems have not been resolved, although they have been slightly modified.
Now QF puts values without data in those regions where it previously put 0. It should be noted that these are not pixels but rather areas much larger than a pixel as seen in the figure. In the case of NC it continues to give NAs in those places and the problem remains that for the L.tiff output some pixels take negative values
I copied access to the files of the new model execution
I really appreciate the help, thanks!
New SWY model
Thanks for trying out 3.14.0 and sending your data along.
The nodata holes are a result of the nodata values found in the soils raster input. Those nodata are now propagated through the model and into QF. So, if you were able to fill in those soils nodata values, that should help with that issue. Even though the LULC and biophysical table have curve number values, it uses the soils raster as a key to determine which column from the biophysical table to apply to the corresponding LULC value.
I actually think this is okay and makes sense for your data. Here is a section of the Users Guide that talks about how local recharge values can be negative.
Precipitation that does not run off as quickflow, and is not evapotranspired by the vegetation on a pixel, can infiltrate the soil to become local recharge. Local recharge can be negative if a pixel does not receive enough of its own water to satisfy its vegetation requirements (determined by its crop factor Kc), so it uses water generated upslope of the pixel as well (referred to as an “upslope subsidy”.)
It looks like the Kc values are high (greater than 1) for some of those LULC where negative values are occurring. So, on top of ETO, you can get more water leaving the pixel from evapotranspiration and vegetation uptake.
Let me know if any of the above is unclear!
@dcdenu4 . Thank you very much for the detailed explanation of the problems. Indeed, the soil map had pixels without data, which coincided with the gaps I saw at the QF output. Thanks for the additional explanation on the curve number!
Regarding the negative values of Local Recharge now I understand it better, in fact I saw that the pattern coincided with the LULC pattern. Which now makes more sense because in the end the Kc associated with this LULC is what determines these values.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.