The invention provides a Yangtze estuary salt marsh wetland macrobenthos habitat simulation prediction method. The method comprises: S1, performing sampling to obtain a sampling sample, and obtainingsampling data; S2, performing zero-value data processing, and obtaining first preprocessing data; S3, obtaining and preprocessing remote sensing data and auxiliary data, and obtaining second preprocessed data; S4, obtaining a training data set and a verification data set; S5, establishing a neural network space model, and training the neural network space model to obtain a trained model; S6, verifying the fitting goodness of the trained model; and S7, screening the trained model to obtain an optimal model, and obtaining a benthic animal prediction habitat map by using the optimal model. According to the Yangtze estuary salt marsh wetland macrobenthos habitat simulation prediction method, a remote sensing means is used, a neural network space model is integrated, the problems of limited benthic organism data and nonlinear environment data of traditional simulation are solved, and the accuracy of macrobenthos community distribution simulation is improved.