Supervised Intelligent Committee Machine Method for Groundwater Level Prediction

Groundwater is an important water resource supplying agricultural, domestic, and industrial needs. Hence, studying and investigation of this vital source is necessary. Over extraction of groundwater cause adverse effects, such as major water level declines [1], consequently water-quality degradation, land subsidence [2], and saltwater intrusion. Therefore, accurate prediction of groundwater level will help planer and managers of water resources and prevent the mentioned effects. The literature shows the success of using artificial intelligence (AI) models in the field of groundwater level such as artificial neural network (ANN) [3], fuzzy logic (FL) [4], and support vector machine (SVM) [5]. Although several studies are reported in the literature that use Supervised committee machine artificial intelligence (SCMAI) as a modeling technique in the field of hydrology e.g. [6] to combine the results of different models to reap the advantages of all AI. This study applied a supervised committee machine with an artificial intelligent (SCMAI) method that replaces linear combination with an artificial neural network. In the SCMAI the ANN receives individual model estimations as input variables and re-predicts the groundwater level.
The Meshginshahr plain is located in Northwest Iran in the Province of Ardabil (Fig. 1). The aquifer of this plain is unconfined. The prevailing climate in this plain is semiarid-cold. The average annual temperature and rainfall are 11.66 ˚C and 292 mm respectively. Sabalan Mount with the height of 4814 meters asml, is the highest point in the study area