作者：
Song, YingYang, TaoLi, ZhenyaXu, Chong-YuHohai Univ
Ctr Global Change & Water Cycle State Key Lab Hydrol Water Resources & Hydraul Eng Nanjing 210098 Peoples R China Hohai Univ
Natl Cooperat Innovat Ctr Water Safety & Hydrosci Nanjing 210098 Peoples R China Chinese Acad Sci
Key Lab Watershed Geog Sci Nanjing Inst Geog & Limnol Nanjing 210008 Peoples R China Univ Oslo
Dept Geosci & Hydrol POB 1047 Blindern N-0316 Oslo Norway
Flow direction algorithms have important application in attracting geomorphic features and topographic attributes, which serve as inputs for some hydrological and topographical models. Evaluating flow direction algori...
详细信息
Flow direction algorithms have important application in attracting geomorphic features and topographic attributes, which serve as inputs for some hydrological and topographical models. Evaluating flow direction algorithms is of great significance and often conducted on synthetic surfaces instead of real digital elevation models for free of approximation errors. However, most widely-used synthetic surfaces are too simplified to represent complex topographical relief of real-world terrains. For this, this work applies a complex synthetic surface of modified Himmelblau's function (HF) to simulate sophisticated terrains encountered in real landscapes. HF surface is spatially smooth and continuous with four hilltops and one valley, where plan curvatures are clustered mainly from -0.1 to 0.1. In addition, a slope line-based discretization numerical (SLDN) approach is designed for obtaining numerical solution to theoretical total contributing area (TCA) on synthetic surfaces of non-integrable slope lines (e.g. HF surface). TCAs estimated by several flow direction algorithms are compared with SLDNderived TCA quantitatively. Results indicate that the largest and smallest mean sizeerrors are obtained by Random eight-node (Rho8) (i.e. 82.6 %) and Freeman multiple flow direction (FMFD) (i.e. 17.9 %), while the largest and smallest mean extenterrors by Eight drainage directions (D8) (i.e. 128.2 %) and Eight drainage directions, least transversal deviation (D8-LTD) (i.e.55.4 %). Most mean errors are larger than 20.0 %, which may not be satisfactory in practice. This work can provide a reference for flow direction algorithms application in digital terrain analysis, and therefore improve accuracy of hydrological, geological and geomorphological models.
暂无评论