Prediction and Analysis of Global Temperature Based on BP and ELMAN Neural Networks

Abstract

In order to explore the global climate evolution and change patterns, this article uses global temperature data from 1881 to 2020 for nearly 140 years, and based on the global temperature zone division model, constructs BP neural network and ELMAN neural network prediction models to analyze the spatiotemporal evolution trend of global temperature historical data. It is found that the average temperature in the northern and southern hemispheres began to significantly increase around 1950; based on the above model, it is predicted that the global annual average temperature will reach its peak around 2050 and continue to maintain around 16.6433℃ for the next fifty years.