应用多元方格网模式比较坡度计算法差异之研究 - 台大地理系.pdfVIP

47-63 (2008) 47 JOURNAL OF GEOGRAPHICAL SCIENCE (51) : 47-63 (2008) Applying a Multifactor Gridded Network Model to Compare the Difference of Gradient Calculating Techniques * Ching-Sheng Chiu Abstract The grid is the basic unit for gradient estimation for both contour-line-base and DEM-base gradient calculating methods. To understand the impact of different gridded networks on computed gradient, the calculating stability and ratio of restricted development (RRD) of different gradient calculating techniques have been tested by the proposed multifactor gridded network model, which generates enormous types of gridded networks by altering grid size, origin shift, and axis rotation. Based on the test which utilizes a 1:1,000 scale topographic map, it is found that the gradient estimated by contour-line-intersection (8L), third-order infinite difference (3X) and weighting third-order infinite difference (3W) is affected by grid size, origin shift, and axis rotation. The percentage of higher gradient and RRD increase as the grid size decreased from 50m to 5m. The test results show that the 8L method has the best calculating stability and highest RRD. Owing to the high variance of gradient determined by altering the axis rotation, there are no obvious clues to conclude the difference among gradient calculating techniques on the RRD.