Deep Learning based Weather Forecast: A Prediction

Abstract

To predict the future weather condition, the probability that the weather on the day of consideration will be least same as the previous day forecast but the chances of it becoming similar in the next two weeks are high. So, processing the weather data of two weeks from the last year slide window is required to choose a size equal to a week. Every quick window week coincides with the current year. Furthermore, the prediction is done based on a window algorithm slide. The results of the method suggest that, the utilization of proposed method to forecast the weather is effective with an average accuracy of 94.2%. Whereas, the radar remote-sensing arena is one of the most exciting and creative future technological enhancements for PWS. Also, the next-generation radar systems (dual-polarization radar, phased-array radar) will enhance the extreme weather detection, rainfall forecasts, and winter weather warnings, and at the same time it will improve the lead time for severe weather threats including tornadoes and heavy rain/flash flood events.