Good afternoon, everyone. I have a problem calculating soil loss in the “Sediment Delivery Ratio” option. It turns out that I’m getting very high soil loss values when I divide the USLE by the area in hectares. For example, I had losses in native forest of 6 Mg/ha and in coffee areas of 22 Mg/ha, this on a flat terrain. I’d like to know what could be happening, please.
Hi @Deri ,
Thanks for writing in to the forum with your question about the Sediment Delivery Ratio (SDR) model.
Which version of InVEST are you using? Please upload your log file to this post so we can take a closer look at what’s happening.
InVEST version 3.14.0 was recently released and it includes a major change to the SDR model in terms of how the LS-factor is calculated, resulting in significantly lower values for USLE and export than previous versions would output. You can read a discussion about the details of these changes on GitHub. If you have not updated to InVEST version 3.14.0 yet, I strongly recommend doing so. Download the latest installers from our site, here. Then, try running the model again and check if the results are more sensible.
Good morning, how are you? I used version 3.13, but I also used 3.14. The RUSLE column in the attributes table represents soil loss per hectare (Mg/ha) and I’m considering the value very high, especially for native forest and coffee, because it’s practically a flat area. The only factor that could perhaps increase these values is erosivity, which for the south of Minas Gerais, Brazil, is above 7000. However, I did similar work in a nearby area and the values are very different. Anyway, if you can help me. Sincerely, thank you. If you like, I can also upload the database I used for InVEST.
Thanks for sharing your log file and the screenshot of your attribute table. Would you please upload your biophysical table so we can examine the
usle_p values? Be sure to include the LULC names that match those in the “Usos” column of your results table. Is each watershed really composed of only one unique LULC class?
Thanks, buddy. The biophysical table is attached.
biophysical_table_Gura2016.csv (197 Bytes)
Thanks, buddy. The biophysical table is attached. In fact, each watershed corresponds to a different class of land use and occupation, so that I can break down soil losses according to class.
2019_Watershead.zip (177.3 KB)
Here is another database, for 2019 year, if you are interested.
DEM, erodibility, rainfall, biophysical table, watershed (land use and occupation) and land use raster.
It appears to me that the USLE per area values you calculated are strongly driven by the
usle_c * usle_p product of those two input parameters. This is especially true considering each catchment contains only a single LULC class:
The higher this product, the more erosion the model predicts. I’m unsure what the class “Facilities” represents exactly, but we typically assign built or urban areas a
usle_c * usle_p of ~
0.99, which is on the other end of the spectrum from your
0.000001. How are you parameterizing these values? What are your sources?
Hi, good morning. In fact, the most important classes for me are coffee and native forest, and in both I obtained the C and P values in the specialized literature for tropical regions. I got the C for coffee from this article for Brazil:
The C of the native forest I obtained from Silva et al. (2016)
Silva, B.C.P., Silva, M.L.N., Batista, P.V.G., Pontes, L.M., Araújo, E.F. De, Curi, N., 2016.
Soil and Water Losses in Eucalyptus Plantation and Natural Forest and Determination
of the USLE Factors at a Pilot Sub-basin in Rio Grande do Sul, Brazil. 40.pp.
Values for facilities and water were desconsidered, cause they can’t influence the soil loss.
Thanks @Deri ,
InVEST results are typically best interpreted as relative values, such as comparing erosion and/or export in one watershed as compared to another or between LULC classes.
To put confidence in the absolute values, it is critical that calibration of parameters is performed to match the realities of your precise area of interest (AOI) as closely as possible. The best way to do this is by comparing results to observed data, like sediment loads in a river or stream within your AOI. Please read the User Guide for details about how to compare modeled SDR results with observations.