Suitability Assessment of Cage Fish Farming Location in Reservoirs through Neural Networks-Based Remote Sensing Analysis

Abstract

The present study evaluates the application of different artificial intelligence methods associated with remote sensing data processing for assessing water quality parameters, with a focus on fish cage farming in the reservoirs. Three AI methods were utilized including 1—optimal artificial neural network (ONN), 2—adaptive neuro fuzzy inference system in which a hybrid algorithm was used for the training process (ANFIS) and 3—coupled evolutionary algorithm-adaptive neuro fuzzy inference system in which particle swarm optimization was utilized in the training process (EA-ANFIS). Three critical water quality parameters for cage fish farming were selected consisting of water temperature, dissolved oxygen (DO) and total dissolved solids (TDS). Moreover, two measurement indices, the Nash–Sutcliffe model efficiency coefficient (NSE) and root mean square error (RMSE), were utilized to assess the predictive skills of the data driven models. Based on the results in the case study, EA-ANFIS is the best method to simulate water temperature and DO in the reservoir by the remote sensing technique. Furthermore, the ANFIS-based model is the best method to simulate TDS. According to the results in the case study, utilizing the spectral images might not be reliable to simulate DO concentration in the reservoirs. However, the images are robust to simulate water temperature as well as TDS concentration.