A Comparison of BPNN, GMDH, and ARIMA for Monthly Rainfall Forecasting Based on Wavelet Packet Decomposition

Abstract

Accurate rainfall forecasting in watersheds is of indispensable importance for predicting streamflow and flash floods. This paper investigates the accuracy of several forecasting technologies based on Wavelet Packet Decomposition (WPD) in monthly rainfall forecasting. First, WPD decomposes the observed monthly rainfall data into several subcomponents. Then, three data-based models, namely Back-propagation Neural Network (BPNN) model, group method of data handing (GMDH) model, and autoregressive integrated moving average (ARIMA) model, are utilized to complete the prediction of the decomposed monthly rainfall series, respectively. Finally, the ensemble prediction result of the model is formulated by summing the outputs of all submodules. Meanwhile, these six models are employed for benchmark comparison to study the prediction performance of these conjunction methods, which are BPNN, WPD-BPNN, GMDH, WPD-GMDH, ARIMA, and WPD-ARIMA models. The paper takes monthly data from Luoning and Zuoyu stations in Luoyang city of China as the case study. The performance of these conjunction methods is tested by four quantitative indexes. Results show that WPD can efficiently improve the forecasting accuracy and the proposed WPD-BPNN model can achieve better prediction results. It is concluded that the hybrid forecast model is a very efficient tool to improve the accuracy of mid- and long-term rainfall forecasting.

Fault Prediction Based on Leakage Current in Contaminated Insulators Using Enhanced Time Series Forecasting Models

“…LSTM has been applied in deep learning by several authors due to its promising features in dealing with non-linear data [58]. GMDH has performance advantages because it is an adaptive model that disregards neurons that do not help in the training process [59].…”

Prediction Model and Influencing Factors of CO2 Micro/Nanobubble Release Based on ARIMA-BPNN

“…An increase in CO 2 concentration in short-distance spaces affects the CO 2 concentration in long-distance spaces over time, owing to environmental factors, such as wind speed, temperature, and humidity. The CO 2 concentration in long-distance spaces reacts with the concentration in short-distance spaces, which is affected by the density of CO 2 gas [32,33]. To describe the temporal correlation of a single measurement point and the spatial correlation of multiple measurement points, a "spatiotemporal coupling coefficient" with spatiotemporal characteristics is proposed to describe the relationship between CO 2 release concentrations in the experimental environment and quantitatively describe the intensity of spatial correlation at different times.…”

Using a Self-Organizing Map to Explore Local Weather Features for Smart Urban Agriculture in Northern Taiwan

“…Thus, the patterns of a large number of clusters and the transitional nodes between patterns can be more readily understood and discerned [60]. The SOM technique preserves the neighborhood relations of the input data to form a meaningful topological map [30] so that a large amount of information can be stored in the weight values of the SOM's neurons with similar characteristics in input vectors [61,62]. An SOM is capable of conserving the space continuum between daily meteorological datum so that there is some resemblance in the neighboring clusters.…”

“…The seasonal pattern of weather types requires probing into years of historical data of various weather factors, and feature patterns can be extracted through effective and efficient data-mining techniques to explore more information that might not otherwise be disclosed. Various machine learning methods were employed in precedent studies to classify weather features and make forecasts [13,14], for example, a deep neural network (DNN) for weather forecasting [15]; a recurrent neural network (RNN) and long short-term memory (LSTM) for air temperature forecasting [16]; an RNN for hourly rainfall forecasting during typhoon periods [17]; a multilayer perceptron neural network (MLP) for air temperature prediction inside greenhouse [18]; a convolutional neural network (CNN) for wind speed prediction [19] and weather pattern clustering [20]; a deep convolutional neural network (DCNN) for weather phenomenon classification based on images [21]; a backpropagation neural network (BPNN) for weather system prediction [22][23][24]; a self-organizing map (SOM) for estimating meteorological variables of evaporation [25], a method to train an SOM for clustering high-dimensional flood inundation maps [26]; an adaptive model of the enhanced multiple linear regression model (EMLRM) for rainfall forecasting [27]; a combined modular models comparison using moving average (MA), MLP, and support vector regression (SVR) for daily and monthly prediction on rainfall time series [28]; an artificial neural network (ANN)-based lower upper bound estimation (LUBE) and multi-objective fully informed particle swarm (MOFIPS) for interval forecasting for streamflow discharge [29]; and a comparison of BPNN, group method of data handing (GMDH), and autoregressive integrated moving average (ARIMA) for monthly rainfall forecasting [30].…”