Reservoir surface temperature prediction method based on space-time bidirectional attention mechanism

Abstract

The invention discloses a reservoir surface temperature prediction method based on a space-time bidirectional attention mechanism, and the method comprises the following steps: S1, data preprocessing; s2, performing time sequence analysis on the LST time sequence obtained after the preprocessing operation in the step S1 in the research area, comparing the time sequence with the temperature of an adjacent meteorological station, and extracting time sequence characteristics; s3, using a PCAN network to extract a microclimate boundary feature map of the research area; and S4, constructing a space-time bidirectional attention prediction model based on LSTM + Attention on the basis of the LSTM, and calculating a prediction result. Compared with BPNN and LSTM prediction, the prediction performance of the prediction method based on the space-time bidirectional attention mechanism provided by the invention is obviously improved; the influence difference of the microclimate effect characteristics on the prediction result is obvious, the prediction result in the same microclimate characteristic is more stable, the variation fluctuation of the prediction result in the microclimate boundary region is large, and the abnormal value deviation is more obvious; and the space attention mechanism has an obvious inhibition effect on different coverage feature boundary abnormal values.

Description

technical field [0001] The invention relates to the technical field of surface temperature prediction, in particular to a method for predicting surface temperature of reservoirs based on a spatio-temporal two-way attention mechanism. Background technique [0002] Land surface temperature (LST) reflects the heat exchange process of surface water, and it and vegetation index are very important in the research of ecological balance and climate change. There are two types of temperature resources in urban areas. The first is the atmospheric temperature calculated based on the meteorological station network, and the second is the surface temperature estimated based on thermal infrared remote sensing technology. Given the complexity of land surface temperatures, surface measurements cannot practically provide values ​​over large areas. With the development of space remote sensing, satellite data provide the only possibility to measure the surface temperature with full space mean ...

AI Technical Summary

Problems solved by technology

Considering the heterogeneity of space, some research methods including spatial regression and geographic weighted regression (GWR) have been proposed. These research results have effectively solved the problems of LST timing gap and spatial heterogeneity, but there is still a long-term dependence of time series relationship problems

Images