USLE and SWAT models are under estimated erosion. So please suggest such empirical based erosion model, so that one can very accurately predict the erosion rate.
Dear nikhil presently machine learning algorithms and R programming can be utilized to do any kind of modelling with very much high accuracy level.
You can go through the below references
1)https://www.mdpi.com/2076-3417/8/8/1369
2)https://www.mdpi.com/2073-4441/11/6/1129
3)https://rd.springer.com/book/10.1007/978-981-10-8043-2
4)https://www.springerprofessional.de/en/gully-erosion-studies-from-india-and-surrounding-regions/17406426?tocPage=1
USLE deals with soil loss from surface erosion. It does not take into account channelized flow such as rills, gullies. I am not that familiar with SWAT. The WEPP model might work, but I have not tried it, and may require more data than some might be willing to collect. One model we tried was the David Rosgen BANCS model. This model requires substantial on site and prework data collection, and we were fortunate to have access to erosion bank study results from NC State University in same physiographic area. Rosgen‘s training sessions and book on sediment describes in more detail. Another approach may be collecting aerial or ground based LiDAR through time, and use change detection to estimate volumes of soil erosion from due to landform and channel change. From our paper on Sediment from a Small Ephemeral Gully (in Researchgate), we found that site erosion occurred in numerous circumstances, but sediment delivery down the ephemeral channel to the collection site was episodic, during severe events. Gully erosion rates depend a lot on the various site, rainfall and runoff factors. Past And ongoing land management can contribute to rates, and as well freeze -thaw events, strong variable winds, etc.
Could you clarify what type of erosion you are planning to simulate, i.e. rill and interrill or gully erosion?
If you are looking for a model to predict gully erosion rates, I have to tell you it does not exist yet. SWAT, WEPP, USLE or any such model will not work as these are incapable of accounting for channelised erosion process. If you are looking to simulate gully growth (without estimate of soil losses due to the growth), it is possible through physically-based modelling. Foster and Lane (1983) or Sidorchuk (1999) are rather well known in this regard. However, in order to employ one of these methods, you would first need high-resolution data on gully geometry, which will most likely involve extensive fieldwork. This paper is a nice example - Article Physically based model for gully simulation: application to ...
I do however disagree with Shubham Roy's answer above, as ML-based approaches would neither give any account of gully evolution nor erosion rates.
The best approach to measure gully erosion rates is using multi-temporal satellite imagery and tracking gully head retreat back in time. I assume (from your previous papers) that you are interested in the permanent gullies of Central India, in which case you would probably need to make use of historical photographs (if any) or satellite imagery to obtain a reliable account of gully growth rates. Even then, field surveys cannot be discounted, as information on gully depth and bulk density of the soil would be necessitated to translate the gully head retreat rates into actual soil erosion rates.
Hope I have been of help!
- Badges
- Science topic
- Similar topics
- Filing