Details
Original language | English |
---|---|
Article number | 178 |
Journal | Earth science informatics |
Volume | 18 |
Issue number | 1 |
Publication status | Published - 15 Jan 2025 |
Abstract
In this research, by proposing an improved metaheuristic algorithm, it will be shown that the results of empirical soil erosion models can be more accurate by integrating with more advanced algorithms that can fill the limitations of previous methods. Modelling Revised Universal Soil Loss Equation (RUSLE) in combination with Sediment delivery ratio (RUSLE_SDR) in Taleghan watershed, located in the north of Iran, is the case study of this research. Two non-convex optimization methods are applied to construct the rainfall-runoff erosivity surface using rainfall data and Digital Elevation Model (DEM) instead of traditional interpolation methods since in fact it involves a non-convex target function due to existence of the random observed quantities which appear in all parts of the model. The first method to find the optimal solution is based on Whale method as a metaheuristic approach. One of the main advantages of the Whale is solving non-convex optimization problems where analytical methods may get stuck in a local minimum. As one may not have perfect information about the corresponding stochastic model, the second method is a new Whale algorithm based on total least square- variance components estimation (TLS-VCE) in which variance matrices are estimated during the non-convex optimization. The rates of soil erosion and sediment yield are estimated by RUSLE_SDR using three methods: the ordinary least-squares (LS) method, the whale algorithm, and the proposed (VCE_whale) algorithm. The modeling results using the three methods are compared with sediment yield measured at the hydrometric station. The results show that the VCE_whale algorithm (with a relative error of 3.12%) performs better than either the whale method (with a relative error of 5.46%) or the LS method (with a relative error of 6.88%). The whale algorithm also performs better than the LS.
Keywords
- Non-convex optimization, RUSLE, Sediment yield, Soil erosion, TLS-VCE, Whale algorithm
ASJC Scopus subject areas
- Earth and Planetary Sciences(all)
- General Earth and Planetary Sciences
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: Earth science informatics, Vol. 18, No. 1, 178, 15.01.2025.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - RUSLE/SDR for assessment of soil erosion by a new non-convex regression
AU - Ebrahimzadeh, Somayeh
AU - Alavipanah, Seyed Kazem
AU - Attarchi, Sara
AU - Mahboub, Vahid
AU - Motagh, Mahdi
N1 - Publisher Copyright: © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2025.
PY - 2025/1/15
Y1 - 2025/1/15
N2 - In this research, by proposing an improved metaheuristic algorithm, it will be shown that the results of empirical soil erosion models can be more accurate by integrating with more advanced algorithms that can fill the limitations of previous methods. Modelling Revised Universal Soil Loss Equation (RUSLE) in combination with Sediment delivery ratio (RUSLE_SDR) in Taleghan watershed, located in the north of Iran, is the case study of this research. Two non-convex optimization methods are applied to construct the rainfall-runoff erosivity surface using rainfall data and Digital Elevation Model (DEM) instead of traditional interpolation methods since in fact it involves a non-convex target function due to existence of the random observed quantities which appear in all parts of the model. The first method to find the optimal solution is based on Whale method as a metaheuristic approach. One of the main advantages of the Whale is solving non-convex optimization problems where analytical methods may get stuck in a local minimum. As one may not have perfect information about the corresponding stochastic model, the second method is a new Whale algorithm based on total least square- variance components estimation (TLS-VCE) in which variance matrices are estimated during the non-convex optimization. The rates of soil erosion and sediment yield are estimated by RUSLE_SDR using three methods: the ordinary least-squares (LS) method, the whale algorithm, and the proposed (VCE_whale) algorithm. The modeling results using the three methods are compared with sediment yield measured at the hydrometric station. The results show that the VCE_whale algorithm (with a relative error of 3.12%) performs better than either the whale method (with a relative error of 5.46%) or the LS method (with a relative error of 6.88%). The whale algorithm also performs better than the LS.
AB - In this research, by proposing an improved metaheuristic algorithm, it will be shown that the results of empirical soil erosion models can be more accurate by integrating with more advanced algorithms that can fill the limitations of previous methods. Modelling Revised Universal Soil Loss Equation (RUSLE) in combination with Sediment delivery ratio (RUSLE_SDR) in Taleghan watershed, located in the north of Iran, is the case study of this research. Two non-convex optimization methods are applied to construct the rainfall-runoff erosivity surface using rainfall data and Digital Elevation Model (DEM) instead of traditional interpolation methods since in fact it involves a non-convex target function due to existence of the random observed quantities which appear in all parts of the model. The first method to find the optimal solution is based on Whale method as a metaheuristic approach. One of the main advantages of the Whale is solving non-convex optimization problems where analytical methods may get stuck in a local minimum. As one may not have perfect information about the corresponding stochastic model, the second method is a new Whale algorithm based on total least square- variance components estimation (TLS-VCE) in which variance matrices are estimated during the non-convex optimization. The rates of soil erosion and sediment yield are estimated by RUSLE_SDR using three methods: the ordinary least-squares (LS) method, the whale algorithm, and the proposed (VCE_whale) algorithm. The modeling results using the three methods are compared with sediment yield measured at the hydrometric station. The results show that the VCE_whale algorithm (with a relative error of 3.12%) performs better than either the whale method (with a relative error of 5.46%) or the LS method (with a relative error of 6.88%). The whale algorithm also performs better than the LS.
KW - Non-convex optimization
KW - RUSLE
KW - Sediment yield
KW - Soil erosion
KW - TLS-VCE
KW - Whale algorithm
UR - http://www.scopus.com/inward/record.url?scp=85217675517&partnerID=8YFLogxK
U2 - 10.1007/s12145-024-01669-w
DO - 10.1007/s12145-024-01669-w
M3 - Article
AN - SCOPUS:85217675517
VL - 18
JO - Earth science informatics
JF - Earth science informatics
SN - 1865-0473
IS - 1
M1 - 178
ER -