Adaptive Determination of the Flow Accumulation Threshold for Extracting Drainage Networks from DEMs

Wei Zhang,Shengjie Zheng,Shuya Liu,Han Zhang,Xiuguo Liu,Wenkai Li,Hugo A Loaiciga

doi:10.3390/rs13112024

Abstract

Selecting the flow accumulation threshold (FAT) plays a central role in extracting drainage networks from Digital Elevation Models (DEMs). This work presents the MR-AP (Multiple Regression and Adaptive Power) method for choosing suitable FAT when extracting drainage from DEMs. This work employs 36 sample sub-basins in Hubei (China) province. Firstly, topography, the normalized difference vegetation index (NDVI), and water storage change are used in building multiple regression models to calculate the drainage length. Power functions are fit to calculate the FAT of each sub-basin. Nine randomly chosen regions served as test sub-basins. The results show that: (1) water storage change and NDVI have high correlation with the drainage length, and the coefficient of determination (R2) ranges between 0.85 and 0.87; (2) the drainage length obtained from the Multiple Regression model using water storage change, NDVI, and topography as influence factors is similar to the actual drainage length, featuring a coefficient of determination (R2) equal to 0.714; (3) the MR-AP method calculates suitable FATs for each sub-basin in Hubei province, with a drainage length error equal to 5.13%. Moreover, drainage network extraction by the MR-AP method mainly depends on the water storage change and the NDVI, thus being consistent with the regional water-resources change.

Highlights

Water is a crucial resource for humans and the environment
It has been argued the optimal threshold is defined by the point at which the second derivative of a power-law fitted to the drainage length as a function of the flow accumulation threshold (FAT) approaches zero
This study proposes method calculateexthe planatory variables for extracting drainage networks, namely, maximum surface roughFAT adaptively, which employs power functions relating the FAT and drainage length in ness, andThis water storage change

Summary

Introduction

Its distribution and flow characteristics have unique geographical traits. Extracting hydrologic information based on DEMs has received substantial attention in the hydrologic analysis of river networks. Classic problems in distributed hydrological modeling are the calculation of flow accumulation and the extraction of drainage networks [1,2]. A key variable in the extraction of drainage networks from DEMs is the flow accumulation threshold (FAT) [3]. The flow accumulation threshold is related to the upslope contributing area at a given point in a drainage area represented by a DEM. The FAT represents the minimum number of upslope cells flowing into a flow-receiving cell that is part of a drainage network (notice the FAT is a dimensionless variable). Applying an FAT to flow accumulation values delineates the drainage network from a DEM

Methods

Results

Discussion

Conclusion