Gross Error Elimination In DEM Data Based On Robust Methods

The widespread availability of Geographic Information System (GIS) datasets and powerful computers, Digital Elevation Model (DEM) as the essential sources of GIS, has raised the quality control problem of input data to be a crucial one. For any DEM project, accuracy, efficiency, and economy are the three main factors to be considered. Accuracy is the most important factor to be considered. If the accuracy of DEM cannot meet the requirements, the whole project needs to be repeated and the efficiency and economy will ultimately be affected. The factors that affect DEM accuracy are various. The presence of gross error will distort the spatial variation present in DEM. In some cases, totally undesirable and unacceptable results may be produced in DEM as well as its products due to the existence of such gross errors. Therefore, the detection and elimination of gross errors in DEM data has become particularly important.In the modern data processing theory, the elimination of gross errors are mainly from two aspects: First, gross errors will be classified as function model to detect gross errors. But this method cannot automatically achieve the position of gross errors. Second, gross errors will be classified as random model to achieve the position of gross errors. Then using statistical methods eliminate gross errors. This method is called Error Theory of Robust Estimation.This article is on this basis. The idea of robust estimation is added to the algorithm of detecting gross errors in DEM , in order to design the model errors, especially with the algorithm of resist errors. They are: Based on the weight iteration of the robust estimation to eliminate gross errors, Based on the weight iteration of variance estimation to eliminate gross errors, and Based on the weight iteration of the initial values to eliminate gross errors. The difference is that: The former is to determine residual errors firstly by least squares adjustment, while the latter is by linear programming method. Based on residual errors and parameters, we calculate the new weight value for each observation according to the weight function. Through iterative calculation, we may obtain residual errors of the observation, then we use the statistical methods to eliminate gross errors. The same experiments prove that all three methods have the good ability of detection and elimination of gross errors, and the results are consistent.