CGRU-based strong space-time characteristic radar echo proximity prediction method

Abstract

The invention discloses a CGRU-based strong space-time characteristic radar echo proximity prediction method, which comprises the following steps: (1) obtaining a continuous radar echo image related to weather proximity prediction, preprocessing the continuous radar echo image, and constructing tensor data with unified time dimension and space dimension; (2) constructing and training a 3DCNN-CGRUnetwork training model to obtain a 3DCNN-CGRU coding prediction network model; (3) inputting the tensor data of the continuous radar echo image sequence for weather proximity prediction in the step (1) into the 3DCNN-CGRU network model to generate a weather proximity prediction result; according to the invention, the 3DCNN-CGRU network model is provided, the transmission capability of spatial-temporal features is enhanced, the spatial-temporal feature correlation of continuous radar echo images is captured and learned more effectively, and the problems that spatial-temporal information is easyto lose and the prediction accuracy is low are solved.

Description

technical field [0001] The invention relates to the technical field of meteorological observation, in particular to a CGRU-based radar echo nowcasting method with strong spatio-temporal characteristics. Background technique [0002] The goal of radar echo nowcasting is to make timely and accurate forecasts of the weather conditions in the local area within a relatively short period of time (for example, 0-2 hours) in the future. At present, this technology has been widely used in residents' travel, agricultural production, flight safety, etc. It can not only facilitate people, but also help disaster prevention and mitigation. With the current climate change and the acceleration of urbanization, the atmospheric conditions have become more and more complex, and various meteorological disasters have occurred frequently. Climate change has brought many negative impacts on people's life and work, and increased many uncertainties. If the above-mentioned meteorological disasters c...

AI Technical Summary

Problems solved by technology

Although the LSTM network with a common structure can solve the meteorological timing problem to a certain extent, there is a strong temporal-spatial correlation in radar echo prediction, and the spatio-temporal information at the previous moment can determine the prediction at the next moment. Spatial correlation, so it is easy to lead to the loss of spatio-temporal information, the decline of prediction accuracy, and the speed cannot be guaranteed

Images