Statistical Forecasting of Vegetation Indices using Integrated Neuro-Fuzzy Inference System with Bio-Inspired Techniques

Abstract

Excess of water usage from limited water resources leads to drought situation in many part of world. Drought analysis and prediction techniques based on time series rainfall and temperature data exists in the literature. The proposed Neuro- Fuzzy Inference System (NFIS) with bio-inspired techniques in this paper for drought prediction related to agriculture production is discussed in this paper. It uses important indices for drought analysis and prediction such as Standardized Precipitation Index (SPI)., Standard Precipitation Evapotranspiration Index (SPEI) and Moisture Adequacy Index (MAI). Three bio-inspired algorithms such as Genetic Algorithm (GA)., Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) have been used with NFIS. The rainfall and temperature dataset is collected from 2002–2018 for Bellary., the district of Karnataka. The error rates for NFIS-GA., NFIS-PSO., NFIS-ACO., and NFIS models are 0.0296., 0.0322., 0.0358 and 0.0456., respectively while the accuracy rates are 1., 0. 9791(1 month was incorrectly predicted out of 48)., 1 and 0.8548 (3 months incorrectly predicted)., respectively. When the accuracy is 1., it indicates that all the months were predicted correctly out of 48 months.