Modeling methods for tidal flat digital terrain based on neural network

The purpose of this paper is to model tidal flat digital terrain. The study tidal flats are in the yellow sea radial sand ridges eastern China. Based on the regularity and variability characteristics of changeable tidal flats, combined with remote sensing and remote surveying technology and information, this research focuses on tidal flat digital terrain modeling by neural network. The model structure is 2 hidden layers, 11 neurons in every layer. The terrain calculated is very close to the terrain actual surveyed. RMSE is 0.1564 m. R
²
is 095817. Residual distribution is normal. The study value is creative to get dynamic tidal flat terrain information fast and efficiently.