Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms
Abstract Land degradation is the cumulative result of several physical and anthropogenic factors combined with adversity in many nations worldwide, particularly in developing nations like India. This study investigates land degradation in lower Gangetic West Bengal, an eastern Indian state, which ha...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | Discover Geoscience |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44288-025-00187-6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849766661817630720 |
|---|---|
| author | Gopal Chowdhury Ashis Kumar Saha |
| author_facet | Gopal Chowdhury Ashis Kumar Saha |
| author_sort | Gopal Chowdhury |
| collection | DOAJ |
| description | Abstract Land degradation is the cumulative result of several physical and anthropogenic factors combined with adversity in many nations worldwide, particularly in developing nations like India. This study investigates land degradation in lower Gangetic West Bengal, an eastern Indian state, which has received limited attention. Two advanced machine learning models were used: the Multilayer Perceptron Neural Network (MLP-NN) and the Radial Basis Function Neural Network (RBF-NN). Traditional approaches, like the Analytical Hierarchy Process (AHP), are prone to subjective bias. Similarly, models like Random Forest and REPTree often fail to capture complex nonlinear relationships. The applied neural networks overcome these drawbacks. A total of 179,916 samples were analyzed. Among them, 25.85 percent were degradation points and 74.15 percent were non-degradation points. Eleven geophysical and environmental variables were selected based on literature and statistical evaluation. Land degradation zones were mapped using field surveys, published reports, and Google Earth imagery. Model performance was assessed using ROC-AUC curves and confusion matrix analysis on a 30 percent test dataset. The MLP-NN achieved an ROC-AUC of 85.20 percent and an accuracy of 88.90 percent. The RBF-NN obtained an ROC-AUC of 84.20 percent and an accuracy of 87.10 percent. Both models highlighted geology, rainfall erosivity, elevation, soil moisture, land use and land cover as key determinants. These results present a reliable, data-driven method for land degradation assessment. The study offers actionable insights for sustainable land management. It also provides crucial policy implications. Regional planners can use these results to design conservation strategies that protect agriculture and improve environmental resilience in lower Gangetic West Bengal. |
| format | Article |
| id | doaj-art-4fbd7034ff49479ba65341f7d899ba89 |
| institution | DOAJ |
| issn | 2948-1589 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Springer |
| record_format | Article |
| series | Discover Geoscience |
| spelling | doaj-art-4fbd7034ff49479ba65341f7d899ba892025-08-20T03:04:30ZengSpringerDiscover Geoscience2948-15892025-07-013112810.1007/s44288-025-00187-6Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithmsGopal Chowdhury0Ashis Kumar Saha1Department of Geography, Delhi School of Economics, University of DelhiDepartment of Geography, Delhi School of Economics, University of DelhiAbstract Land degradation is the cumulative result of several physical and anthropogenic factors combined with adversity in many nations worldwide, particularly in developing nations like India. This study investigates land degradation in lower Gangetic West Bengal, an eastern Indian state, which has received limited attention. Two advanced machine learning models were used: the Multilayer Perceptron Neural Network (MLP-NN) and the Radial Basis Function Neural Network (RBF-NN). Traditional approaches, like the Analytical Hierarchy Process (AHP), are prone to subjective bias. Similarly, models like Random Forest and REPTree often fail to capture complex nonlinear relationships. The applied neural networks overcome these drawbacks. A total of 179,916 samples were analyzed. Among them, 25.85 percent were degradation points and 74.15 percent were non-degradation points. Eleven geophysical and environmental variables were selected based on literature and statistical evaluation. Land degradation zones were mapped using field surveys, published reports, and Google Earth imagery. Model performance was assessed using ROC-AUC curves and confusion matrix analysis on a 30 percent test dataset. The MLP-NN achieved an ROC-AUC of 85.20 percent and an accuracy of 88.90 percent. The RBF-NN obtained an ROC-AUC of 84.20 percent and an accuracy of 87.10 percent. Both models highlighted geology, rainfall erosivity, elevation, soil moisture, land use and land cover as key determinants. These results present a reliable, data-driven method for land degradation assessment. The study offers actionable insights for sustainable land management. It also provides crucial policy implications. Regional planners can use these results to design conservation strategies that protect agriculture and improve environmental resilience in lower Gangetic West Bengal.https://doi.org/10.1007/s44288-025-00187-6Lower Gangetic West BengalLand degradationMLP-NNRBF-NNROC-AUC |
| spellingShingle | Gopal Chowdhury Ashis Kumar Saha Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms Discover Geoscience Lower Gangetic West Bengal Land degradation MLP-NN RBF-NN ROC-AUC |
| title | Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms |
| title_full | Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms |
| title_fullStr | Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms |
| title_full_unstemmed | Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms |
| title_short | Assessing land degradation in lower gangetic west bengal using GIS-based soft computing and advanced machine learning algorithms |
| title_sort | assessing land degradation in lower gangetic west bengal using gis based soft computing and advanced machine learning algorithms |
| topic | Lower Gangetic West Bengal Land degradation MLP-NN RBF-NN ROC-AUC |
| url | https://doi.org/10.1007/s44288-025-00187-6 |
| work_keys_str_mv | AT gopalchowdhury assessinglanddegradationinlowergangeticwestbengalusinggisbasedsoftcomputingandadvancedmachinelearningalgorithms AT ashiskumarsaha assessinglanddegradationinlowergangeticwestbengalusinggisbasedsoftcomputingandadvancedmachinelearningalgorithms |