Impact of Input Feature Selection on Groundwater Level Prediction From a Multi-Layer Perceptron Neural Network

Abstract

With the growing use of machine learning (ML) techniques in hydrological applications, there is a need to analyze the robustness, performance, and reliability of predictions made with these ML models. In this paper we analyze the accuracy and variability of groundwater level predictions obtained from a Multilayer Perceptron (MLP) model with optimized hyperparameters for different amounts and types of available training data. The MLP model is trained on point observations of features like groundwater levels, temperature, precipitation, and river flow in various combinations, for different periods and temporal resolutions. We analyze the sensitivity of the MLP predictions at three different test locations in California, United States and derive recommendations for training features to obtain accurate predictions. We show that the use of all available features and data for training the MLP does not necessarily ensure the best predictive performance at all locations. More specifically, river flow and precipitation data are important training features for some, but not all locations. However, we find that predictions made with MLPs that are trained solely on temperature and historical groundwater level measurements as features, without additional hydrological information, are unreliable at all locations.

Long-Term Missing Value Imputation for Time Series Data Using Deep Neural Networks

“…The dataset used in the study consisted of approximately eight years (2010-2018) of daily observations. For a more detailed description of the dataset and the data sources, we refer the reader to [22,34]. Figure 1 shows the time series plots for all variables.…”

Predicting Changes in Spatiotemporal Groundwater Storage Through the Integration of Multi-Satellite Data and Deep Learning Models

“…Hyperparameters are variables used in deep learning models, and they are generally based on the prior knowledge of the experimenter about the optimal implementation of deep learning models [51]. Deep learning is a complex type of neural network with various hyperparameters such as batch size, training iterations, and hidden layers that should be tuned before training.…”

Application of Machine Learning Models to Predict Maximum Event Water Fractions in Streamflow

“…In recent years, machine learning algorithms have been widely employed for efficient simulations of high dimensional and nonlinear relationships of various hydrological variables in surface and subsurface hydrology. They have been employed to predict streamflow (Wu and Chau, 2010;Rasouli et al, 2012;Senthil Kumar et al, 2013;He et al, 2014;Shortridge et al, 2016;Abdollahi et al, 2017;Singh et al, 2018;Yuan et al, 2018;Adnan et al, 2019bAdnan et al, , 2021bDuan et al, 2020), groundwater and lake water level (Yoon et al, 2011;Tapoglou et al, 2014;Li et al, 2016;Sahoo et al, 2017;Sattari et al, 2018;Malekzadeh et al, 2019;Sahu et al, 2020;Yaseen et al, 2020;Kardan Moghaddam et al, 2021), water quality parameters such as nitrogen, phosphorus, and dissolved oxygen (Chen et al, 2010;Singh et al, 2011;Liu and Lu, 2014;Kisi and Parmar, 2016;Granata et al, 2017;Sajedi-Hosseini et al, 2018;Najah Ahmed et al, 2019;Knoll et al, 2020), soil hydraulic conductivity (Agyare et al, 2007;Das et al, 2012;Elbisy, 2015;Sihag, 2018;Araya and Ghezzehei, 2019;Adnan et al, 2021a), soil moisture (Gill et al, 2006;Ahmad et al, 2010;Coopersmith et al, 2014;…”

“…They found that the LSTM neural network performed better than the other ANN models. Sahu et al (2020) analyzed the accuracy and variability of groundwater level predictions obtained from a MLP model with optimized hyperparameters for different amounts and types of available training data. Yaseen et al (2020) developed a MLP with Whale optimization algorithm for Lake water level forecasting.…”