Prediction of remotely sensed cloud related parameters over an inland urban city of India: a neuro-computing approach

Abstract

Artificial neural network (ANN) is a mathematical model useful for forecasting on the any type of available data. This tool is not only useful in environment but also covers wide ranges of applicability. Utilizing this model, a study was carried out in an inland area of Nagpur for forecasting satellite-derived cloud parameters. Nine ANN architects are developed based on five pollutant parameter (aerosol optical depth, RSPM, SPM, SO2, NOx), meteorological and some cloud parameter. The models are used to simulate concentration of pollutants as well as the forecast and validation of cloud top temperature, cloud ice water path and cloud liquid water path during different seasons (winter, pre-monsoon and post-monsoon). Models based on back-propagation neural network were tested using the collected data of study area. The ANN models were trained using gradient descent algorithms to check the robustness and adaptability of the models. ANN models based on both satellite and ground-based data variables demonstrate the best performance and are skilled at resolving patterns of pollutant dispersion to the atmosphere during 2006–2013 for Nagpur city.