I have tested the InVEST SDR model in many watersheds of the Ethiopian Highlands and the performance of the model is very good for smaller watersheds. In the larger watersheds, the model overestimates soil loss. I have checked why the model does this and finally, I understand that it is from the LS equation.
Therefore, my recommendation to the developer of the model is the following:
- To allow the user to use its own LS factor like what can we do for the R and K factors
- Otherwise, include at least four or three common LS factor equations. So that the user can choose the best one.
2 Likes
Hi @Temesgen, thank you for your thoughts. Indeed, there are various approaches to calculate LS (Wischmeier and Smith, Desmet and Govers, etc). However, while they result in different LS values, I think it is difficult to decide which one is better, especially if there are no data for validation, which is the case for many users.
Likely it is better to stay with one LS factor calculation to ensure consistency of results. However, it might be possible to give users more control over the LS factor calculation, e.g., the maximum L value (now set to 333m) or similar parameters.
Out of curiosity, I would also be interested to know which tools you would usually use to calculate LS (or would you do it from scratch in a GIS)?
Best and thanks for your input.
Rafael
3 Likes
Thank you. I am using raster calculator in ArcGIS to drive the LS factor. I think, allowing users to use their own LS factor equation is important besides to the available LS algorithm.
Hi @Temesgen, thanks for the suggestion. We will discuss if that can be a feature in a future release. As a workaround, you could use the USLE results, which are basically K X C X P X LS and divide it by the LS raster from InVEST. That will give you the values for K X C X P. Then multiply with your LS raster and you will get USLE again but with your LS. You can also multiply results with SDR to get to sediment export. The main thing you would not get is information on spatial patterns of sediment retention.
1 Like
Dear Rafael Schmitt
Thank you for your nice suggestion. In addition to the LS factor algorithm, the SDR sub-model also requires some correction on the P factor. The current version of the model allows users to assign P factors for the corresponding land-use types. That means it assumes that the level of implemented soil and water conservation practices in each land use type is uniform. However, in reality, it may not be true. Assume that one is working on a basin or watershed with an area of 10,000km2. Do you think that the same conservation measures are present, for example, in cultivated land? Mostly, we do not get the same level of conservation. Thus, digitalizing the soil and conservation (SWC) measures and preparing the P factor by itself is required. In this case, a user-defined P factor is needed in the InVEST SDR model. In general, for wider applications of the SDR model, user-defined LS and P factors are important besides the available options. I think, most people in the future need to use the InVEST SDR model than SWAT (Soil and Water Assessment Tool) particularly in data-scarce regions of the world. However, correction of the above-mentioned issues are key.
With Kind Regards
Temesgen Gashaw (PhD)
Assistant Professor of Environmental Science
Department of Natural Resource Management
College of Agriculture and Environmental Science
Bahir Dar University, Ethiopia
https://scholar.google.com/citations?hl=en&user=pmRQh5UAAAAJ
1 Like
Dear Temesgen, thank you for your suggestions and use cases. We will look into those. Regarding your question about the P factor: I agree there will be many cases where you have improved soil conservation in some places, but not in others, within the same watershed or study area.
For now, you could work your way around it by defining a different landuse class in the land-use-land-cover raster, and assign a different P factor to it in the biophysical table. E.g., define one land use class as āagricultureā and one as āagriculture with improved practicesā. Each gets a different land use code and a separate row in the biophysical table.
1 Like
Dear Rafael Schmitt
Thank you for your suggestions.
Regards
Temesgen