Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques
This study highlights the importance of sustainable land management in preserving soil health and agricultural productivity, particularly in mitigating land degradation. Soil Quality Index (SQI) was assessed in Egypt’s Nile River Basin using 266 surface samples (0–30 cm depth) collected between 2021...
| Autores principales: | , , |
|---|---|
| Formato: | Journal Article |
| Lenguaje: | Inglés |
| Publicado: |
Elsevier
2025
|
| Materias: | |
| Acceso en línea: | https://hdl.handle.net/10568/175482 |
| _version_ | 1855516145125163008 |
|---|---|
| author | Singha, Chiranjit Sahoo, Satiprasad Govind, Ajit |
| author_browse | Govind, Ajit Sahoo, Satiprasad Singha, Chiranjit |
| author_facet | Singha, Chiranjit Sahoo, Satiprasad Govind, Ajit |
| author_sort | Singha, Chiranjit |
| collection | Repository of Agricultural Research Outputs (CGSpace) |
| description | This study highlights the importance of sustainable land management in preserving soil health and agricultural productivity, particularly in mitigating land degradation. Soil Quality Index (SQI) was assessed in Egypt’s Nile River Basin using 266 surface samples (0–30 cm depth) collected between 2021 and 2022. Eleven key soil quality indicators such as bulk density (BD), sand, silt, clay, pH, electrical conductivity (EC), organic carbon (OC), calcium (Ca), nitrogen (N), phosphorus (P), and potassium (K) were analyzed to estimate the observed SQI (SQIobs) using a PCA-based scoring method and geostatistical techniques. The SQIobs were validated against in-situ wheat yield. Various hybrid stacking ensemble (SE) machine learning models including Random Forest (SE-RF), Extreme Gradient Boosting (SE-XGB), Gradient Boosting Machine (SE-GBM), Multivariate Adaptive Regression Splines (SE-MARS), Support Vector Machine (SE-SVM), and SE-Cubist was applied to predict soil quality (SQIpred) in data-scarce regions. The SE-RF and SE-Cubist models demonstrated the highest predictive accuracy (R2 = 0.830 and 0.824, respectively). Results showed that “very high” and “very low” SQI classes covered 24.25 % and 14.70 % of the study area, respectively. Future projections using CMIP6 models indicate a decline in SQI, from 24.25 % to 19.15 % (SSP2-4.5) and 10.85 % (SSP5-8.5) between 1990 and 2030. SHAP analysis identified BD, clay, sand, OC, and N as key drivers of SQIobs, while SM, Tmax, FC, ST, and NDVI significantly influenced SQIpred. This study provides a robust framework for assessing soil quality, offering valuable insights for land use planning, sustainable agriculture, and combating soil degradation. |
| format | Journal Article |
| id | CGSpace175482 |
| institution | CGIAR Consortium |
| language | Inglés |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Elsevier |
| publisherStr | Elsevier |
| record_format | dspace |
| spelling | CGSpace1754822026-01-14T06:13:41Z Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques Singha, Chiranjit Sahoo, Satiprasad Govind, Ajit nile river machine learning wheat pca wheat yield soil quality index (sqi) This study highlights the importance of sustainable land management in preserving soil health and agricultural productivity, particularly in mitigating land degradation. Soil Quality Index (SQI) was assessed in Egypt’s Nile River Basin using 266 surface samples (0–30 cm depth) collected between 2021 and 2022. Eleven key soil quality indicators such as bulk density (BD), sand, silt, clay, pH, electrical conductivity (EC), organic carbon (OC), calcium (Ca), nitrogen (N), phosphorus (P), and potassium (K) were analyzed to estimate the observed SQI (SQIobs) using a PCA-based scoring method and geostatistical techniques. The SQIobs were validated against in-situ wheat yield. Various hybrid stacking ensemble (SE) machine learning models including Random Forest (SE-RF), Extreme Gradient Boosting (SE-XGB), Gradient Boosting Machine (SE-GBM), Multivariate Adaptive Regression Splines (SE-MARS), Support Vector Machine (SE-SVM), and SE-Cubist was applied to predict soil quality (SQIpred) in data-scarce regions. The SE-RF and SE-Cubist models demonstrated the highest predictive accuracy (R2 = 0.830 and 0.824, respectively). Results showed that “very high” and “very low” SQI classes covered 24.25 % and 14.70 % of the study area, respectively. Future projections using CMIP6 models indicate a decline in SQI, from 24.25 % to 19.15 % (SSP2-4.5) and 10.85 % (SSP5-8.5) between 1990 and 2030. SHAP analysis identified BD, clay, sand, OC, and N as key drivers of SQIobs, while SM, Tmax, FC, ST, and NDVI significantly influenced SQIpred. This study provides a robust framework for assessing soil quality, offering valuable insights for land use planning, sustainable agriculture, and combating soil degradation. 2025-06-01 2025-07-03T18:58:42Z 2025-07-03T18:58:42Z Journal Article https://hdl.handle.net/10568/175482 en Limited Access Elsevier C. Singha, S. Sahoo and A. Govind, Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques, Advances in Space Research, https://doi.org/10.1016/j.asr.2025.03.058 |
| spellingShingle | nile river machine learning wheat pca wheat yield soil quality index (sqi) Singha, Chiranjit Sahoo, Satiprasad Govind, Ajit Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title | Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title_full | Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title_fullStr | Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title_full_unstemmed | Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title_short | Transforming soil quality index predictions in the Nile River Basin using hybrid stacking machine learning techniques |
| title_sort | transforming soil quality index predictions in the nile river basin using hybrid stacking machine learning techniques |
| topic | nile river machine learning wheat pca wheat yield soil quality index (sqi) |
| url | https://hdl.handle.net/10568/175482 |
| work_keys_str_mv | AT singhachiranjit transformingsoilqualityindexpredictionsinthenileriverbasinusinghybridstackingmachinelearningtechniques AT sahoosatiprasad transformingsoilqualityindexpredictionsinthenileriverbasinusinghybridstackingmachinelearningtechniques AT govindajit transformingsoilqualityindexpredictionsinthenileriverbasinusinghybridstackingmachinelearningtechniques |