Research On NDVI Prediction Of Haixi Prefecture Using Graph Neural Network And Time Series Decomposition Algorithm

Normalized Differential Vegetation Index(NDVI)is a commonly used index in remote sensing.It reflects the distribution and change of vegetation coverage and is of great significance to the monitoring of vegetation change,rational use of land resources and other ecological environment-related fields.Basing on the graph neural network,this paper proposes a decomposition prediction model of NDVI——STL-GNN.This model can not only accurately predict NDVI,but also achieve the purpose of multi-site synchronous prediction.To verify the effectiveness of the proposed model,this paper takes Haixi Prefecture,China as the research object and predicts NDVI value of six areas of the prefecture.The length of prediction is one year.Meantime,the paper selects eight mixed models and four single models as comparison.The prediction results show the model is better than other contrast models,which verify the effectiveness of the proposed model.In addition,sequence decomposition method has been widely used in time series prediction,and previous studies lack of comprehensive comparison among sequence decomposition algorithms.In this paper,graph neural network is used as the prediction model to comprehensively compare the differences among the five widely used decomposition algorithms,which provides a reference for the future use of decomposition algorithms.