GWLM–NARX. Grey Wolf Levenberg–Marquardt-based neural network for rainfall prediction
| DOI | https://doi.org/10.1108/DTA-08-2019-0130 |
| Published date | 08 January 2020 |
| Pages | 85-102 |
| Date | 08 January 2020 |
| Author | Razeef Mohd,Muheet Ahmed Butt,Majid Zaman Baba |
| Subject Matter | Library & information science,Librarianship/library management,Library technology,Information behaviour & retrieval,Metadata,Information & knowledge management,Information & communications technology,Internet |
GWLM–NARX
Grey Wolf Levenberg–Marquardt-based neural
network for rainfall prediction
Razeef Mohd, Muheet Ahmed Butt and Majid Zaman Baba
University of Kashmir, Hazratbal, India
Abstract
Purpose –Weather forecasting is the trending topic around the world as it is the way to predict the threats posed
by extreme rainfall conditions that lead to damage the human life and properties. These issues can be managed
only when the occurrence of the worse weather is predicted in advance, and sufficient warnings can be executed in
time. Thus, keeping in mind the importance of the rainfall prediction system, thepurpose of this paper is to propose
an effective rainfall prediction model using the nonlinear auto-regressive with external input (NARX) model.
Design/methodology/approach –The paper proposes a rainfall prediction model using the time-series
prediction that is enabled using the NARX model. The time-series prediction ensures the effective prediction
of the rainfall in a particular area or the locality based on the rainfall data in the previous term or month or
year. The proposed NARX model serves as an adaptive prediction model, for which the rainfall data of the
previous period is the input, and the optimal computation is based on the proposed algorithm. The adaptive
prediction using the proposed algorithm is exhibited in the NARX, and the proposed algorithm is developed
based on the Grey Wolf Optimization and the Levenberg–Marqueret (LM) algorithm. The proposed algorithm
inherits the advantages of both the algorithms with better computational time and accuracy.
Findings –The analysis using two databases enables the better understanding of the proposed rainfall
detection methods and proves the effectiveness of the proposed prediction method. The effectiveness of the
proposed method is enhanced and the accuracy is found to be better compared with the other existing
methods and the mean square error and percentage root mean square difference of the proposed method are
found to be around 0.0093 and 0.207.
Originality/value –The rainfall prediction is enabled adaptively using the proposed Grey Wolf Levenberg–
Marquardt (GWLM)-based NARX, wherein an algorithm, named GWLM, is proposed by the integration of
Grey Wolf Optimizer and LM algorithm.
Keywords Neural network, GWO, Grey Wolf Optimizer, Levenberg–Marquardt algorithm, NARX model,
Rainfall prediction
Paper type Research paper
1. Introduction
Rainfall isthe major rigorous phenomena in a climate system, which has a directinfluence on
ecosystems, agriculture and water resource management. The deep rainfall causes floods,
mudslides, landslides and other natural disasters. The rainfall-induced disasters lead to severe
losses and damage to both infrastructure and life. In the Indian economy, agriculture is the
backbone, and most of the agriculture depends on the rainfall (Hirani and Mishra, 2016).
Rainfall prediction is a problem, which requires to be solved to minimize or avoid losses of life
and properties. Moreover, rainfall prediction (Ramu et al.,2017;Yuet al., 2015) possesses
considerable significance in the modelling of the runoff that brings the management of the
water resources. The severity of the rainfall is predicted based on the accurate and quantitative
prediction, and moreover, it enables to forecast the flash flooding in addition to the details of
the hydrologic interests. The prediction of the rainfall is significantly needed in the areas that
are subjected to the flash flooding and in these places, the prediction accuracy depends on the
frequency of the rainfall in that particular area, and the prediction could be made in advance
(Altunkaynak and Nigussie, 2015). Accurate rainfall prediction (Mohammad et al., 2017a, b;
Shah et al., 2017a, b) is the major concern in most of the countries. Thus, accurate rainfall Data Technologies and
Applications
Vol. 54 No. 1, 2020
pp. 85-102
© Emerald PublishingLimited
2514-9288
DOI 10.1108/DTA-08-2019-0130
Received 10 August 2019
Revised 19 October 2019
Accepted 6 December 2019
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/2514-9288.htm
This paper forms part of a special section: “Knowledge and data mining for recent and advanced
applications using emerging technologies”.
85
GWLM based
neural network
prediction plays a beneficial role in designing the policies, planning and final decisions, and
above all, it assures a sustainable functioning of water resource systems (Cramer et al., 2017;
Wadoux et al., 2017; Kashiwao et al., 2017; Adil et al., 2017; Mohammad et al., 2017a, b).
Modelling the rainfall prediction is a tedious process because the process of modelling
injects complexity as the processes of the atmosphere seems to be complex (Sedki et al.,
2009; Teschl et al., 2007; Wu and Jin, 2009; Wu et al., 2010, 2015). The accurate prediction of
the rainfall enhances the quality of decision making and the corresponding activities (Huang
and Davy, 2016; Thorey et al., 2015; Shah et al., 2017a, b). A broad range of rainfall prediction
techniques is employed at national and regional levels. In general, rainfall prediction
techniques are classified into two types, such as empirical methods and dynamical methods.
The empirical method depends on the analysis of past historical data of weather (Azad and
Dixit, 2019a, b). The mainly utilized empirical methods for rainfall prediction includes the
auto-regressive integrated moving average, artificial neural networks (ANNs) (Vijaya et al.,
2016; Shah et al., 2017a), multiple linear regression, K-nearest neighbours and support vector
machines for regression methods (Ninu Preetha and Praveena, 2018; Adil et al., 2017). These
types of algorithms possess the tendency to evaluate multiple points simultaneously in the
search space and also, they are capable of determining the global solution for a given
problem (Mohammad et al., 2017a, b). Moreover, it employs a simple scalar performance
measure that does not employ and utilize derivative information (Wu et al., 2015). In the
dynamical method, predictions are created by physical models depends on the system of
equations, which predict future rainfall. The weather forecasting by computer using
equations is called numerical weather prediction (NWP) (Hirani and Mishra, 2016).
Predicting the weather accurately is a hectic challenge, and the forecasters are often
blamed if there is any variation in the forecast (Lee and Liu, 2004; Innocenti et al., 2017;
Wichitarapongsakun et al., 2016; Singh et al., 2017). The major challenges in the weather
forecast are regarding the prediction techniques that are based on the NWP. NWP is not
effective in detecting weather type, and wind scale that constitute the errors in short-term
daily reference evapotranspiration (ETo) forecasts as a result of the change in the climate
zone (Yang et al., 2016). The drawback of the linear regression model is regarding the
resolution corresponding to the NWP and the measurements signify the local effects that
minimize the variability of solar power generation when energy (More and Ingle, 2017)
sources are spatially aggregated (Huang and Davy, 2016). The performance of the linear
regression model (Huang and Davy, 2016) is found to be less with the increase in the
forecast horizon of the weather. The performance of the linear regression model reduces
further during the variation of cloud type. The contiguous rain area (CRA) technique
(Sagar et al., 2017) offers huge errors regarding the shape/structure of the rainstorm
events. The clusterwise linear regression (CLR) is tough to undergo informed policy,
planning and management decisions in the operation of water resources systems
(Adil et al., 2017).
Accurate forecasting of rainfall hasbeen one of the most important issues in hydrological
research because early warnings of severe weather can help prevent casualties and damages
caused by natural disasters, if timely and accurately forecasted. This paper proposes a
rainfall prediction model using the time-series prediction that is enabled usingthe Grey Wolf
Levenberg–Marquardt–nonlinearauto-regressive withexternal input (GWLM–NARX) model.
The time-seriesprediction ensures the effectiveprediction of the rainfall in a particular area or
the locality based on the rainfall data in the previous term or month or year. The proposed
GWLM-NARXmodel serves as an adaptive prediction model, for whichthe rainfall data of the
previous period is the input, and the optimal computation is based on the proposed GWLM
algorithm. The adaptive prediction using the proposed GWLM algorithm is exhibited in the
NARX, and the proposed algorithmis developed based on the Grey WolfOptimization (GWO)
and the Levenberg–Marquardt (LM) algorithm. The proposed GWLM-NARX model is
86
DTA
54,1
To continue reading
Request your trial