Comparison of Evolving Connectionist Systems (ECoS) and Neural Networks for Modelling Daily Pan Evaporation from Algerian Dam Reservoirs

Abstract

Evaporation (EP) from dams’ reservoirs measured using pans is one of the most important methods adopted for quantifying the loss of water through evaporation. Black box artificial intelligence techniques (AI) have been developed as alternative approaches for quantifying evaporation, and several kinds of models have been proposed worldwide. The present study uses the measurement of several climatic variables such as air temperature, wind speed, and relative humidity to test the performances of new AI techniques called evolving connectionist systems (ECoS), applied for predicting daily evaporation from several dam reservoirs located in Algeria country. Two ECoS models, namely, (1) offline-based dynamic evolving neural-fuzzy inference systems named DENFIS_OF and (2) online-based dynamic evolving neural-fuzzy inference systems named DENFIS_ON, were applied and compared for predicting daily evaporation. The results using ECoS models were compared to multiple linear regression (MLR) and artificial neural network (ANN) models. From the results obtained, it is seen that the ECoS models could predict daily evaporation from dam reservoirs with better accuracy than the ANN and MLR models.