Simulation of Titicaca Lake Water Level Fluctuations Using Hybrid Machine Learning Technique Integrated with Grey Wolf Optimizer Algorithm

Abstract

Lakes have an important role in storing water for drinking, producing hydroelectric power, and environmental, agricultural, and industrial uses. In order to optimize the use of lakes, precise prediction of the lake water level (LWL) is a main issue in water resources management. Due to the existence of nonlinear relations, uncertainty, and characteristics of the time series variables, the exact prediction of the lake water level is difficult. In this study the hybrid support vector regression (SVR) and the grey wolf algorithm (GWO) are used to predict lake water level fluctuations. Also, three types of data preprocessing methods, namely Principal component analysis, Random forest, and Relief algorithm were used for finding the best input variables for prediction LWL by the SVR and SVR-GWO models. Before the LWL simulation on monthly time step using the hybrid model, an evolutionary approach based on different monthly lags was conducted for determining the best mask of the input variables. Results showed that based on the random forest method, the best scenario of the inputs was Xt−1, Xt−2, Xt−3, Xt−4 for the SVR-GWO model. Also, the performance of the SVR-GWO model indicated that it could simulate the LWL with acceptable accuracy (with RMSE = 0.08 m, MAE = 0.06 m, and R2 = 0.96).

Bio-Inspired Hybridization of Artificial Neural Networks: An Application for Mapping the Spatial Distribution of Soil Texture Fractions

“…However, this model has several considerable shortcomings, such as inconsistent architectures for different applications, coupled with the process required to tune and fit a neural network, which is a time-consuming procedure that is largely based on trial and error [27,28]. Conventionally, ANNs have been fitted using a backpropagation (BP) algorithm; however, state-of-the-art approaches using bio-inspired, metaheuristic, optimization algorithms have become increasingly prevalent, including the genetic algorithm (GA) [29], particle swarm optimization (PSO) [30], ant lion optimization (ALO) [31], spotted hyena optimizer (SHO) [32], binary spring search algorithm (BSSA) [33], grey wolf algorithm (GWO) [34], genetic optimization resampling based particle filtering (GORPF) algorithm [35], and ant colony optimization (ACO) [16]-all of which may be hybridized with ANNs to address the aforementioned disadvantages.…”

A review on the applications of machine learning for runoff modeling

“…In hydrological modeling studies, accurate runoff modeling is the main research topic that affects water resources planning, including dam design, water resource allocation plans, catchment area management, and flood management (Nourani et al 2009;Zhou et al 2019;Chadalawada et al 2020;Mohammadi et al 2020b). It is scientifically proven that due to the physical processes and natural changes related to the river system, then the prediction of the river system and its runoff behavior is particularly difficult to analyze.…”

Community perspectives on conservation of water sources in Tarkeshwar sacred groves, Himalaya, India

“…There is increasing awareness in society that habitat degradation, overexploitation, climate change and pollution are major threats to biodiversity and ecosystems across the globe (Ayvaz & Elci 2012;Pereira et al 2012;Mohammadi et al 2020). Ecosystem degradation and biodiversity loss undermine the functioning of ecosystems and their ability to provide goods and services, which are of fundamental importance for the well-being of present and future generations of people world-wide (de Groot et al 2002;Sinha & Mishra 2015).…”