Abstract
Lakes have a crucial role in the industrial, agricultural, environment, and drinking water fields. Accurate prediction of lake levels is one of the most important parameters in the reservoir management and lakeshore structure designing. The goal of the present study is to examine the robustness of two different Genetic Algorithm-based regression methods namely the Genetic Algorithm Artificial neural network (GAA) and the Genetic Programming (GP) by considering their performance in predicting the non-observed lakes. To do that, data collected from the four-year daily measurements of the Chahnimeh#1 lake in Eastern Iran were used for developing the GAA and GP models and after that, the performance of the considered models are examined to predict the lake water levels of an adjacent lake namely Chahnimeh#4 as the non-observed information. The results showed that both model has the ability to simulate adjacent lakes using the considered lake water levels for the training procedure. In addition, another goal is to develop simple, practical formulation for predicting the lake water level, So that, using the GP method, as the superior model, three different formulations are proposed in order to predict the one, three, and five days ahead lake water level, respectively.
Acknowledgment
The authors express their sincere thanks from Regional Water Authority of Sistan Baluchestan for the Lake water level observance dataset.