Integration of Wavelet Transform with ANN and WNN for Time Series Forecasting: an Application to Indian Monsoon Rainfall
- PDF / 452,112 Bytes
- 5 Pages / 595.276 x 790.866 pts Page_size
- 57 Downloads / 179 Views
SHORT COMMUNICATION
Integration of Wavelet Transform with ANN and WNN for Time Series Forecasting: an Application to Indian Monsoon Rainfall Mrinmoy Ray1 • K. N. Singh1 • V. Ramasubramanian1 Santosha Rathod3
•
Ranjit Kumar Paul1
•
Anirban Mukherjee2
•
Received: 20 November 2017 / Revised: 22 May 2018 / Accepted: 9 January 2020 The National Academy of Sciences, India 2020
Abstract Indian agricultural activity relies on monsoon rainfall; hence, its forecasting is indispensable for proper planning. Artificial neural network (ANN) is one of the popular approaches for rainfall forecasting. Recent research activity demonstrates that consolidating diverse model/techniques improves the exactness of forecasting when contrasted with the individual models. Therefore, the present study proposed a hybrid forecasting framework for rainfall forecasting combining wavelet transform, ANN and wavelet neural network (WNN). As a case study, Indian monsoon rainfall time series data have been considered to assess the forecasting performance of the proposed forecasting framework. The proposed approach has been compared with ANN and WNN. Observational outcomes uncover that the forecasting accuracy of the proposed strategy is superior to ANN and WNN. Keywords Wavelet transform ANN WNN Rainfall forecasting
The summer monsoon during the months of June to September is the major rainfall season (70% of its yearly rainfall) for India. Fluctuations in monsoon rainfall influence agriculture, drinking water, energy sector and & Mrinmoy Ray [email protected]; [email protected] 1
ICAR-Indian Agricultural Statistics Research Institute, New Delhi 110012, India
2
ICAR Research Complex for Eastern Region, Patna 800014, India
3
ICAR-Indian Institute of Rice Research, Hyderabad 500030, India
livelihood of a huge number of individuals living in the nation. In particular, Indian agricultural performance depends to a huge degree on the quantum and dispersion of rainfall during these monsoon months. Consequently, forecasting Indian summer monsoon rainfall (ISMR) is very important as its high changeability has noteworthy effect on economy of the nation. Monthly rainfall (in mm) data of India for 4 months, i.e., June till September during the years 1871–2011, accessible at the site ( www.tropmet.res.in) of the Indian Institute of Tropical Meteorology, Pune, India, are considered. Data from 1871 to 1999 were utilized for model development and from 2000 to 2011 were utilized to check the forecasting performance. Till date, ANN has been frequently utilized for forecasting rainfall. The ANN has been employed in several fascinating research studies, to cite a few, monsoon rainfall for the Districts and Sub-Division Kerala, India [1], monsoon rainfall of India [2], rainfall of Bangkok, Thailand [3], rainfall of Chennai, India [4]. Artificial neural networks (ANNs) model are considered as a class of generalized nonlinear models that can catch different nonlinear structures present in the data set. The principle preferred standpoint of t
Data Loading...
