|
Modeling and Evaluating the Performance of Groundwater Resources Drought Index Using XGBoost and SVM Algorithms in Ghorveh-Dehgolan Plain in Kurdistan Province, Iran
|
Ebrahim Yousefi Mobarhan *   |
|
|
|
Abstract: (23 Views) |
The Ghorveh-Dehgolan plain is the largest plain in Kurdistan province, which is of great importance for this province in terms of agriculture, and knowledge of the status of groundwater resources and optimal management of the plain is a detailed study of groundwater level fluctuations. The aim of this research is to investigate and apply machine models to improve the accuracy of predictions and promote groundwater resource management in the Ghorveh-Dehgolan plain, where the Groundwater Resources Index (GRI) is a reliable indicator for monitoring the drought status of the study area during the period. It was used from 1379 to 1399. The results show that the value of the GRI index has a decreasing trend and in 1397, a high degree of hydrogeological drought occurred in the groundwater resources of this region, which in the continuation of this trend will be accompanied by a serious crisis of groundwater level reduction and the consequences for the region. Predictions using machine learning models show that both machine learning models (support vector machine and boost) fit well with the experimental data of the groundwater resource drought index. Validation of the models used in predicting this index using Taylor plot showed that the boost intensity model (XGBoost) with correlation coefficient (r=0.93) and root mean square error (RMSE=0.071) had the best performance in predicting the GRI index as well as the machine learning model. Support vector with correlation coefficient (r=0.87) and RMSE=0.149 showed high performance in predicting the index. The results of this study confirm that machine learning models are suitable for groundwater resource drought prediction indices, which are recommended in other similar areas.
|
Article number: 5 |
| Keywords: Groundwater, GRI, Drought Index, Taylor diagram, Machine learning |
|
|
|
|
Type of Study: Research |
Subject:
Special
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
