An interpolation method for lack of DEM data area in tidal creeks based on neural network

This paper researches an interpolation method for lack of LiDAR DEM data area in tidal creeks. The study area is tidal flats in the yellow sea radial sand ridges eastern China. Based on a large of tidal creeks surveying data, combined with topography and geomorphology laws, this research focuses on an interpolation method for lack of LiDAR DEM data area in tidal creek by neural network. The interpolation model structure is 2 hidden layers, 6 neurons in every layer. The calculated terrain of tidal creek that is lack of DEM data is very similar to the actual surveyed terrain. RMSE is 0.117m. R² is 0.716. Residual distribution is normal. The study value is creative to repair the terrain where is lack of LiDAR DEM in tidal flats.