Bayesian Regularized Neural Network Model Development for Predicting Daily Rainfall from Sea Level Pressure Data: Investigation on Solving Complex Hydrology Problem

Lu Ye,Musaddak M Abdul Zahra,Mou Leong Tan,Saadya Fahad Jabbar,Zaher Mundher Yaseen

doi:10.1155/2021/6631564

Abstract

Prediction of daily rainfall is important for flood forecasting, reservoir operation, and many other hydrological applications. The artificial intelligence (AI) algorithm is generally used for stochastic forecasting rainfall which is not capable to simulate unseen extreme rainfall events which become common due to climate change. A new model is developed in this study for prediction of daily rainfall for different lead times based on sea level pressure (SLP) which is physically related to rainfall on land and thus able to predict unseen rainfall events. Daily rainfall of east coast of Peninsular Malaysia (PM) was predicted using SLP data over the climate domain. Five advanced AI algorithms such as extreme learning machine (ELM), Bayesian regularized neural networks (BRNNs), Bayesian additive regression trees (BART), extreme gradient boosting (xgBoost), and hybrid neural fuzzy inference system (HNFIS) were used considering the complex relationship of rainfall with sea level pressure. Principle components of SLP domain correlated with daily rainfall were used as predictors. The results revealed that the efficacy of AI models is predicting daily rainfall one day before. The relative performance of the models revealed the higher performance of BRNN with normalized root mean square error (NRMSE) of 0.678 compared with HNFIS (NRMSE = 0.708), BART (NRMSE = 0.784), xgBoost (NRMSE = 0.803), and ELM (NRMSE = 0.915). Visual inspection of predicted rainfall during model validation using density-scatter plot and other novel ways of visual comparison revealed the ability of BRNN to predict daily rainfall one day before reliably.

Highlights

Rainfall decides agricultural activities, ecology, and environment of a region. erefore, it is considered as the key factor of social and economic development of any region [1]
Five advanced artificial intelligence (AI) algorithms such as extreme learning machine (ELM), Bayesian regularized neural networks (BRNNs), Bayesian additive regression trees (BART), EGB, and hybrid neural fuzzy inference system (HNFIS) were used considering the complex relationship of rainfall with sea level pressure (SLP). e SLPs over the climate domain having a significant correlation with daily rainfall of East Coast of Peninsular Malaysia (ECPM) were used to compute their principal components and selection of inputs. e APHRODITE rainfall of all the grid points over the ECPM was average to prepare the daily rainfall of the area. e Machine learning (ML) models were used for the prediction of areal averaged daily rainfall of the region. e description of the ML models used in this study is given in the following subsections
E correlation of rainfall with 1- to 3-lag days is presented in three maps in the figure. e colour ramp in the maps is used to show the positive correlation. e significant correlation (p < 0.05) is presented with dots. e figure shows that rainfall in ECPM is positively correlated with SLP in the north and negatively correlated with SLP in the south. e differences in SLP cause movement of air over the region. e moist air from the sea when enters the land causes rainfall. erefore, SLP data can be used for prediction of rainfall in ECPM. e SLP data of nearly 2321 to 2486 are found to correlate with rainfall for different lags

Summary

Introduction

Ecology, and environment of a region. erefore, it is considered as the key factor of social and economic development of any region [1]. The rapid changes in rainfall that earth experienced in recent years and projected for the future cannot be forecasted by empirical models accurately as the models were not developed with such large variability in data [17]. Chen and Sun [21] used the physical-empirical model for performance improvement of a dynamical model in seasonal rainfall forecasting using sea level pressure (SLP) data. Erefore, the development of ML-based physical-empirical forecasting models has grown very fast in recent years. To the best knowledge of the current study, this is the first attempt to use an array of sophistical ML algorithms for the development of physical-empirical models for forecasting daily rainfall of PM. APHRODITE daily rainfall and NCEP ERA daily SLP data for the period 1951−2015 were collected from the corresponding websites

Description of Employed Machine Learning Algorithms

Results and Discussion

Conclusion