High rainfall event identification using remote sensing satellite data integrated with a hybrid deep learning framework

Rainfall estimation is of paramount importance, pertaining to the plethora of stakeholders that place substantial reliance on it, particularly in the agricultural sector. Because of the dynamic character of rainfall, forecasting is challenging and byzantine. In this study, estimation of daily rainfall for an Assamese town, Silonijan located in the Karbi Anglong district of Assam India is conducted. The data used for this research are the openly available gridded products provided by the Indian Meteorological Department (IMD). The objective of this study is to estimate the potential of Artificially Intelligent (AI) techniques in rainfall estimation. This work aims to develop new methods of rainfall prediction by utilizing Deep Learning rather than typical statistical and machine learning approaches, which lack the accuracy depicted by modern hybrid Deep-Learning-based techniques. A hybrid Deep Learning model built by stacking a Gated Recurrent Unit and Bi-Directional Long Short-Term Memory (Bi LSTM) is proposed in this research. The predictions are computed as a result of a multivariate time series analysis problem. Dynamic elements of the environment, maximum and minimum temperatures are used as input parameters for the model, with daily rainfall serving as an output parameter. The data spans over several decades, ranging from January 1950–May 2020. Predictions for a period of 15 days of the month June 2020 are made. The developed model is validated using two robust quality evaluation criteria, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The MAE value of the proposed methodology is 6.97, and the RMSE value is 10.13. Furthermore, the presented model's predictions are confirmed by looking at the actual rainfall received by the Silonijan over a 15-day period in June 2020. Upon comprehensive analysis, it is ascertained that the proposed artificial intelligence model provides a precise estimation of the average rainfall anticipated for the town of Silojinan during the month of June 2020. This deep learning-based research could aid in the accurate and timely forecast of rainfall in places prone to flooding or drought, thus helping in disaster management.