The Study About The Influence Of Error's Spatial Autocorrelation On Terrain Parameters Based On Digital Elevation Model

The Digital Elevation Model (DEM) has error or uncertainty inherently as one of the main spatial databases of Geographical Information System (GIS). The error of DEM will be transmitted and magnified during digital terrain analysis. The error resulted from digital terrain analysis has an important effect on interpretation and explanation which people make when simulating geographical process, and the result of the digital analysis. Since people acknowledge the existence of spatial autocorrelation of alternative, it will be the "worst-case-scenario" if the spatial autocorrelation of the error is not concerned during the spatial analysis. One of the aims of the study was the investigation of the identity of the spatial autocorrelation of the error. Firstly, it gave out three traditional models of spatial autocorrelation of the error which are the Spherical model, the Exponential model and the Gaussian model based on the purpose above. Then this paper researched the discrepancy about these three models and the effect of the main parameters of the model created which are the sill and the range on the spatial autocorrelation of the error.This paper chose three basic terrain parameters which are the slope, the aspect and the contour curvature for digital terrain analysis. It is clear that the key step of the algorithms about the slope, the aspect and the contour curvature is the calculation of p and q which are the gradients at W-E and N-S directions. This paper deduced the accuracy estimating models of the slope algorithm and the aspect algorithm based on the error is spatial autocorrelation and three spatial autocorrelation models theoretically. It also analyzed the influencing rule of the Root Mean Square Error (RMSE) of the DEM and the spatial autocorrelation distance on the accuracy of the slope algorithm and the aspect algorithm.Monte Carlo stochastic simulation technique was used to establish the stochastic error of the experimental sample regions. The stochastic error would have steady spatial autocorrelation after interpolation according to Spherical model, Exponential model and Gaussian model. The aim DEM is gained from plus the error with spatial autocorrelation to the DEM of sample regions.This paper made a statistics of the RMSE of the terrain parameters which were the slope, the aspect and the contour curvature according to the different relief-form feature points which the peak point, the nek point, the ridge point, the valley point, the flat point and the slope-feet point. The result of the experiment consisted with the academic deduction and pointed out the influencing rule of spatial autocorrelation of the error on digital terrain analysis. And it is proved that it is necessary and important to take the error spatial autocorrelation into account when analyzing digital terrain.