Performance Evaluation of Rainfall Spatial Interpolation in Lower Vellar Watershed Using Inelastic Constitutive Artificial Neural Networks

Abstract

Rainfall patterns vary widely, making it challenging to get accurate measurements because rain gauges only provide limited and scattered data points. This challenge is particularly significant when estimating and simulating climate change models within diverse environmental settings, as accurate rainfall data is essential for various scientific applications like hydrological modeling and agricultural planning. In this manuscript, Performance Evaluation of Rainfall Spatial Interpolation in Lower Vellar Watershed Using Inelastic Constitutive Artificial Neural Networks (SIR-LVW-ICANN) is proposed. In this proposed approach data are collected from Weather data Indian cities (1990 to 2022). The input data is fed to pre-processing using High Accuracy Distributed Kalman Filtering (HADKF) to find Missing Data and Normalization. Then, pre-processed data are fed to Inelastic Constitutive Artificial Neural Networks (ICANN) to effectively predict the mean annual precipitation. Then the proposed SIR-LVW-ICANN is implemented in Python and the performance metrics like Accuracy, RMSE and MAE are analysed. Performance of the SIR-LVW-ICANN approach attains 19.36%, 26.42% and 23.27% higher accuracy and 22.36%, 15.42% and 18.27% lower RMSE when analysed through existing techniques like Deep Spatial Interpolation of Rain Field for U.K. Satellite Networks (DSI-RF-CNN) , Application of multiple spatial interpolation approaches to annual rainfall data in the Wadi Cliff basin (MSI-ARD-MLPNN), and Machine Learning Procedures for Daily Interpolation of Rainfall in Navarre (DIR-KNN) methods respectively.