Application of the novel state-of-the-art soft computing techniques for groundwater potential assessment
- Publisher:
- Springer Nature
- Publication Type:
- Journal Article
- Citation:
- Arabian Journal of Geosciences, 2022, 15, (10), pp. 929
- Issue Date:
- 2022-05-05
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| s12517-021-09005-y.pdf | Published version | 10.91 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
Groundwater is one of the important ecological elements which are now considered as vulnerable resources. The absence and shortage of this precious resource can be grown into an ecologically fragile condition. So, identification of this nature of groundwater resources can be useful for proper concern about this resource and its associated measures. The accurate and meaningful prediction of groundwater potential is a very important need for management of water resources in arid and semi-arid regions of the world. Though others have modeled predictions of groundwater distribution using various statistical and machine learning methods, this study tested alternative decision tree (ADTree) and the credal decision tree (CDT) as standalone models as well as in ensembles with dagging, bagging, and decorate. Eighteen groundwater potential conditioning factors (lithology, convergence index, drainage density, elevation, distance to fault, fault density, height above nearest drainage (HAND), distance to surface, train surface texture, topographical wetness index (TWI), land use/land cover, stream transport index (STI), topographical position index (TPI), multi-resolution index of valley bottom flatness (MRVBF), profile curvature (PrC), plan curvature (PC), slope angle, and rainfall) were measured and compiled for 188 spring locations and 188 non-spring locations in the Tabriz Plain of East Azerbaijan Province, Iran. The conditioning factors were tested for multi-collinearity and there was none. The conditioning factors were examined and weighted for their importance for predicting groundwater potential and the data were modeled. The models were trained using 70% of the database and tested with the data for the remaining 30% of the locations. The results indicate that the six ensembles for both decision tree models surpassed the standalone decision trees in terms of success and accuracy. Running the training dataset, the models’ success rates were 72% for ADTree, 90% for ADTree-
Please use this identifier to cite or link to this item: