Support vector regression integrated with novel meta-heuristic algorithms for meteorological drought prediction

• Published in: Meteorology and atmospheric physics Vol. 133; no. 3; pp. 891 - 909
• Main Authors: Malik, Anurag, Tikhamarine, Yazid, Souag-Gamane, Doudja, Rai, Priya, Sammen, Saad Shauket, Kisi, Ozgur
• Format: Journal Article
• Language: English
• Published: Vienna Springer Vienna 01-06-2021; Springer Nature B.V
• Subjects: Agricultural ecosystems; Agriculture; Algorithms; Aquatic Pollution; Atmospheric Sciences; Autocorrelation; Autocorrelation function; Autocorrelation functions; Drought; Drought index; Earth and Environmental Science; Earth Sciences; Heuristic; Heuristic methods; Hydrologic data; Hydrology; Inspection; Math. Appl. in Environmental Science; Meteorology; Mitigation; Monthly rainfall; Monthly rainfall data; Original Paper; Predictions; Problem solving; Rain; Rainfall; Rainfall data; Regions; Statistical analysis; Support vector machines; Terrestrial Pollution; Waste Water Technology; Water Management; Water Pollution Control; Water resources
• ISSN: 0177-7971; 1436-5065
• DOI: 10.1007/s00703-021-00787-0

Drought is a complex natural phenomenon, so, precise prediction of drought is an effective mitigation tool for measuring the negative consequences on agriculture, ecosystems, hydrology, and water resources. The purpose of this research was to explore the potential capability of support vector regression (SVR) integrated with two meta-heuristic algorithms i.e., Grey Wolf Optimizer (GWO), and Spotted Hyena Optimizer (SHO), for meteorological drought (MD) prediction by utilizing EDI (effective drought index). For this objective, the two-hybrid SVR–GWO, and SVR–SHO models were constructed at Kumaon and Garhwal regions of Uttarakhand State (India). The EDI was computed in both study regions by using monthly rainfall data series to calibrate and validate the advanced hybrid SVR models. The autocorrelation function (ACF) and partial-ACF (PACF) were utilized to determine the optimal inputs (antecedent EDI) for EDI prediction. The results produced by the hybrid SVR models were compared with the calculated (observed) values by employing the statistical indicators and through graphical inspection. A comparison of results demonstrates that the hybrid SVR–GWO model outperformed to the SVR–SHO models for all study stations located in Kumaon and Garhwal regions. Also, the results highlighted the better suitability, supremacy, and convergence behavior of meta-heuristic algorithms (i.e., GWO and SHO) for meteorological drought prediction in the study regions.