A novel multiple flow direction algorithm for computing the topographic wetness index

• Published in: Hydrology Research Vol. 43; no. 1-2; pp. 135 - 145
• Main Authors: Yong, Bin, Ren, Li-Liang, Hong, Yang, Gourley, Jonathan J., Chen, Xi, Zhang, You-Jing, Yang, Xiao-Li, Zhang, Zeng-Xin, Wang, Wei-Guang
• Format: Journal Article
• Language: English
• Published: London IWA Publishing 01.01.2012
• Subjects: Algorithms; Catchment area; Digital Elevation Models; Direction; Distribution; Exponents; Hydrologic models; Hydrologic processes; Hydrology; Mathematical analysis; Mathematical models; Moisture content; Rivers; Saturation; Soil; Soil moisture; Spatial distribution; Terrain; Topography; Topography (geology); Watersheds; Wetness; Wetness index
• ISSN: 0029-1277; 1998-9563; 2224-7955
• DOI: 10.2166/nh.2011.115

The topographic wetness index (TWI), frequently used in approximately characterizing the spatial distribution of soil moisture and surface saturation within a watershed, has been widely applied in topography-related geographical processes and hydrological models. However, it is still questionable whether the current algorithms of TWI can adequately model the spatial distribution of topographic characteristics. Based upon the widely-used multiple flow direction approach (MFD), a novel MFD algorithm (NMFD) is proposed for improving the TWI derivation using a Digital Elevation Model (DEM) in this study. Compared with MFD, NMFD improves the mathematical equations of the contributing area and more precisely calculates the effective contour length. Additionally, a varying exponent strategy is adopted to dynamically determine the downslope flow-partition exponent. Finally, a flow-direction tracking method is employed to address grid cells in flat terrain. The NMFD algorithm is first applied to a catchment located upstream of the Hanjiang River in China to demonstrate its accuracy and improvements. Then NMFD is quantitatively evaluated by using four types of artificial mathematical surfaces. The results indicate that the error generated by NMFD is generally lower than that computed by MFD, and NMFD is able to more accurately represent the hydrological similarity of watersheds.