An objective approach for feature extraction: distribution analysis and statistical descriptors for scale choice and channel network identification

摘要

A statistical approach to LiDAR derived topographic attributes for theautomatic extraction of channel network and for the choice of the scale toapply for parameter evaluation is presented in this paper. The basis of thisapproach is to use distribution analysis and statistical descriptors toidentify channels where terrain geometry denotes significant convergences.Two case study areas with different morphology and degree of organizationare used with their 1 m LiDAR Digital Terrain Models (DTMs). Topographicattribute maps (curvature and openness) for various window sizes are derivedfrom the DTMs in order to detect surface convergences. A statisticalanalysis on value distributions considering each window size is carried outfor the choice of the optimum kernel. We propose a three-step method toextract the network based (a) on the normalization and overlapping ofopenness and minimum curvature to highlight the more likely surfaceconvergences, (b) a weighting of the upslope area according to thesenormalized maps to identify drainage flow paths and flow accumulationconsistent with terrain geometry, (c) the standard score normalization ofthe weighted upslope area and the use of standard score values as nonsubjective threshold for channel network identification. As a final step foroptimal definition and representation of the whole network, anoise-filtering and connection procedure is applied. The advantage of theproposed methodology, and the efficiency and accurate localization ofextracted features are demonstrated using LiDAR data of two different areasand comparing both extractions with field surveyed networks.