A new algorithm for the automatic extraction of valley floor width

Abstract

The width of the valley floor dictates the structure and function of the associated river or stream at the macro level. Consequently, the valley floor width is increasingly recognized as a key parameter and has been employed for many applications (e.g., river classification). Here, we propose an automated algorithm approach to extract the valley floor width of U- and V-shaped valleys by using the breakpoint of the slope between the valley wall and valley floor to support digital elevation models (DEMs) of different resolutions. The algorithm includes five steps: 1) preprocessing data, 2) extracting linear reach, 3) extracting perpendiculars of reaches, 4) delineating the outlines of the valley floor and 5) calculating the width of the valley floor. Eight case studies were used to validate the algorithm. Valley floor outlines and widths were computed and measured, revealing that the algorithm performs well with a high correlation coefficient (CC > 0.88) and coefficient of determination (R2 > 0.77), as well as low relative bias (|RB| < 10%). The algorithm performs better when applied in U-shaped valleys than in V-shaped valleys. The algorithm offers a fast and inexpensive tool with which to obtain valley shape information for research and management.