资源环境科技发展态势分析平台

D8-LTD is a path-based flow direction algorithm for estimating total contributing area (TCA). However, the precision of D8-LTD TCA is seriously challenged by the incorporation of flow paths and basic hypothesis of allowing only one receiving pixel. To solve above problems, this work proposes an improved D8-LTD by adopting the strategies of path independency and local dispersion (i.e. LTD-PI-LD). Two indicators are adopted to quantify size and extent precisions of estimated TCAs relative to theoretical TCA. Then, LTD-PI-LD is compared with several typical single-flow-direction (SFD) algorithms (i.e. D8, D8-LTD, LTD-PI) and multiple-flow-direction (MFD) algorithms (i.e. MFD-Quinn, D∞) on synthetic surfaces. Results indicate that (1) TCA maps of D8 and D8-LTD contain too many singular high-TCA tubes and source pixels, which are caused by the incorporation of flow paths and converging characteristics of SFD algorithms. (2) In MFD algorithms, the problem of artificial dispersion triggers the mechanisms of cross compensation and mutual amplification, leading to high size precision but low extent precision in estimated TCAs. (3) Strategy of path independency prevents the occurrence of high-TCA tubes and effectively reduces the number of source pixels. (4) Both size and extent precisions of D8-LTD TCAs are largely improved by introducing the strategies of path independency and local dispersion. Of all algorithms, LTD-PI-LD obtains the highest/second-highest size precision and the highest extent precision in estimated TCAs. As a whole, LTD-PI-LD shows advantages of high precision in estimated TCAs, low computation cost, high efficiency and robustness to DEM (Digital elevation model) resolution and terrain type.

This article is protected by copyright. All rights reserved.