Study On Different Interpolation Algorithms To Generate DEM And The Influence Of Terrain Factor

Currently, Digital Elevation Model (DEM) is widely used in various fields, which is cheap and practical to obtain through digitized topographic maps. Because of the purpose of the user, data sources and landform type, the construction of the DEM showed a certain degree of differences. However, there is not a general interpolation algorithm that can meet practical needs. Hence, it is important to choose suitable parameters of interpolation algorithm and the contrast analysis of DEM derived terrain factor built by different algorithm was analyzed in this paper. With the increase of datas, the construction speed of the DEM relatively slow down accordingly. Through the purchase of the high-performance machine to increase the speed of data processing, it can’t meet the actual demand.With the development of GPU technology, Compute Unified Device Architecture (CUDA)) with unique acceleration effect was used for accelerating the DEM generation speed.Three regions with typical landforms in Fujian Province were choosed as the research areas in this paper. The accuracy, speed and quality of the derived DEM were analyzed and the commonly used interpolation algorithm parameters and errors were comparatively analyzed. We generated the DEM parallelly by the CUDA technology.The results showed that the speed of DEM generation was acceleration obviously. The main content of this paper were as follows:(1) Parameters effects evaluation of interpolation algorithm of DEM generation.The feature of error change in different interpolation algorithm parameters was discussed to find out the parameters for high precision DEM generation. At the same time, for different interpolation algorithms DEM error characteristics were analyzed. The results showed that the generation algorithm anudem, thin plate spline interpolation and Irregular triangulation network had higher accuracy, but IDW, Kriging had larger relative error.(2) Algorithm design and performance elevation of parallel DEM generation.The parallel algorithm of IDW, Spline and Kriging were designed to generate DEM using Block and dynamic search method based on CUDA Multi-Threading. It showed that each algorithm with different amount of data presented certain times acceleration through the comparison the performance of three kind of algorithm.(3) Study on the impact of DEM terrain factor by different interpolation method. The diversity of extracting terrain factor by different interpolation algorithm was explained based on the mathematical statistics and information theory. The suitable situation of each algorithm was realized. There were not flat areas which extract the aspect of the terrain factors from ANUDEM algorithm.This study can provide a reference for common users for data processing and digital analysis using interpolation method. It also can further use of Componet Object Model component technology to package the interpolation algorithm for other researchers to use. The results given in this paper have certain theoretical and practical significance.