Research And Experiment About Gross Error Detection In DEM Data Considering Data Source

As an important expression of the surface topography, the Digital Elevation Model has been greatly concerned by people. The factors that affect the precision of DEM are various. Among these factors, gross error that caused by misplay will distort the space character of DEM, and even caused the distortion of the DEM products. Therefore, the detection of gross errors and avoiding them have been very important.The efficient method detecting gross error in DEM should proceed in the expression of DEM, such as grid DEM and irregular DEM, and they rarely take the characteristic of DEM into account, which based on different data source. All the existing methods acquiring the DEM data have their own advantages and disadvantages, and the characteristics of DEM that produced by those methods are also different. Just depending on analyzing these, for example, DEM produced by airborne laser radar data, scanned and vectorized topographic maps data and multi-source, searching the algorithms, respectively gross error detection algorithm in DEM based on LIDAR data source, gross error detection algorithm in DEM based on scanned and vectorized topographic maps data source and gross error detection algorithm in DEM based on multi-source.1. DEM produced by LIDAR data source, has high spatial resolution, greater density of data point and uniform distribution, and its residual non-ground points existing by the way of cluster after filtering, but it's difficult to extract topographic characteristic line automatically. So gross error detection algorithm in DEM based on LIDAR data source, look the residual non-ground points after filtering as errors, for these errors presented in DEM by cluster, then using error cluster algorithm detect errors. After that, the residual error points are nearly most discrete form, using inverse distance weighted method to calculate the residuals of all points and statistical tests to detect residual error points.2. DEM based on scanned and vectorized topographic maps data source, is a method of routine measurement, it has small density of data point and a high integration of the terrain, achievements of automatically extracted topographic characteristic line been abundance. Now, algorithms of gross error are two available. One is, gross error being classified as function model to detect gross error; another is, gross error being classified as random model to detect gross error. The algorithm of gross error detection in DEM based on scanned and vectorized topographic maps data source, which is classified gross error as random model, is robust. First, it is to determine residual errors by least squares adjustment, then calculate the new weight value for each observation according to the weight function obtains the final residuals of observations through iterative calculation, and last we use the statistical methods to eliminate gross error.3. The algorithm of gross error detection in DEM based on multi-source, because of laser scan with blindness, we cannot collect data points in the topographic characteristic place as manual measurement and be difficult to automatically extract topographic characteristic line; scanned and vectorized topographic maps, has high integration of the terrain and achievements of automatically extracted topographic characteristic line are abundance. So directly extract topographic characteristic line from the topographic map overlapped to the LIDAR data, it can improve the gross error detection rate in DEM.The three algorithms have been tested, and the experimental results show that DEM based on different data source used different algorithms that can enhance the effectiveness of gross error detection in DEM and improve the precision of DEM, and provide some references for the further DEM production.