GIS-based bivariate statistical techniques for groundwater potential analysis (an example of Iran)

Abstract

Groundwater potential analysis prepares better comprehension of hydrological settings of different regions. This study shows the potency of two GIS-based data driven bivariate techniques namely statistical index (SI) and Dempster-Shafer theory (DST) to analyze groundwater potential in Broujerd region of Iran. The research was done using 11 groundwater conditioning factors and 496 spring positions. Based on the ground water potential maps (GPMs) of SI and DST methods, 24.22% and 23.74% of the study area is covered by poor zone of groundwater potential, and 43.93% and 36.3% of Broujerd region is covered by good and very good potential zones, respectively. The validation of outcomes displayed that area under the curve (AUC) of SI and DST techniques are 81.23% and 79.41%, respectively, which shows SI method has slightly a better performance than the DST technique. Therefore, SI and DST methods are advantageous to analyze groundwater capacity and scrutinize the complicated relation between groundwater occurrence and groundwater conditioning factors, which permits investigation of both systemic and stochastic uncertainty. Finally, it can be realized that these techniques are very beneficial for groundwater potential analyzing and can be practical for water-resource management experts.

Optimized Conditioning Factors Using Machine Learning Techniques for Groundwater Potential Mapping

“…Evidently, wells, springs, rivers, etc. are the point of discharge for groundwater from the aquifer layer [2,17]; hence, in this study, spring locations are identified as the groundwater resource indicators. An inventory map from comprehensive field surveys by the Iranian Department of Water Resource Management, Iran (Figure 1), pinpoints the locations of 917 springs.…”

“…Groundwater is more reliable in terms of quality and availability since it exists in many geological formations and is naturally protected from direct contamination [12][13][14]. Moreover, according to the authors of [15][16][17], safe and sustainable water supply is only guaranteed through adequate groundwater resource assessment and planning. Haghizadeh et al [17] adds that proper GPM requires reliable information concerning procedures and associated conditioning factors.…”

“…Moreover, according to the authors of [15][16][17], safe and sustainable water supply is only guaranteed through adequate groundwater resource assessment and planning. Haghizadeh et al [17] adds that proper GPM requires reliable information concerning procedures and associated conditioning factors. In this context, some components, such as dataset(s), computer algorithms, and size of the area under investigation greatly affect the reliability of GPM results [6].…”

“…In recent years, many models have been developed for GPM based on selected indicators. For example, Haghizadeh et al [17] utilized spring locations in two bivariate techniques (i.e., Dempster-Shafer theory and Statistical Index) to analyze potential groundwater zones. Other recent works used machine learning, namely K-nearest Neighbors (KNN) [21], Linear Discriminant Analysis (LDA) [22], multivariate adaptive regression splines [23], quadratic discriminant analysis [21], Support Vector Machine (SVM) [24], Random Forest (RF) [23] and Decision Trees [25].…”

Application of Probabilistic and Machine Learning Models for Groundwater Potentiality Mapping in Damghan Sedimentary Plain, Iran

“…In present the decades, groundwater recharge level has fallen due to unnecessary use and unscientific management plans [9]. Hence, aquifer potential determination through groundwater potentiality analysis is a good strategy in this field [10,11]. Different methods and models, techniques and processes were induced and used for the groundwater potentiality map (GWPM) or identification of areas having good potentiality of groundwater recharge.…”