Short-Term Rainfall Forecasting Aided by IoT and Deep Learning

The frequent increase in extreme weather events makes precipitation forecasting crucial for disaster mitigation. Internet of Things (IoT) and Artificial Intelligence (AI) has been successfully applied in several areas. This paper analyzes AI techniques to predict short-term precipitation volume from data collected by a low cost weather monitoring system. The architecture proposed for the monitoring stations is based on cheaper IoT devices. The IA algorithms applied use intraday data, collected every 10 minutes, from meteorological stations deployed in Recife-PE. We used sensors for precipitation, temperature, air humidity, wind speed and wind gust. The following methods were considered: Linear Regression (LR), Random Forest (RF) Regressor, support vector regression (SVR) and long short-term memory (LSTM). According to our results, the LSTM model presented the best results, obtaining a MAE of 0.03 and an RMSE of 0.04.