Open AccessIssues in Replication and Stability of Least-cost Path Calculations
Author(s) -
Irmela Herzog
Publication year - 2022
Publication title -
studies in digital heritage
Language(s) - English
Resource type - Journals
ISSN - 2574-1748
DOI - 10.14434/sdh.v5i2.33796
Subject(s) - digital elevation model , replication (statistics) , terrain , elevation (ballistics) , computer science , data set , set (abstract data type) , stability (learning theory) , path (computing) , computation , function (biology) , algorithm , geology , data mining , mathematics , geography , statistics , artificial intelligence , remote sensing , cartography , geometry , machine learning , evolutionary biology , biology , programming language
An important and frequently used tool in archaeological spatial analysis is least-cost path (LCP) analysis to compute routes connecting a set of targets. The outcome depends on the cost model chosen and the topographic data used. A slope-dependent cost model requires a digital elevation model (DEM) that should reflect the landscape in the past. It is often impossible to reconstruct the past terrain, and modern high-resolution elevation data results in problematic storage requirements and computation times. This article presents a case study that explores issues in replication and stability of LCP calculations for pairs of targets that are close to known old trade routes. A large number of cost models is tested based on two topographic data sets, including DEMs of two different resolutions (25 m and 50 m). The cost models use six different slope-dependent cost functions suggested by various authors for pedestrian movement. Moreover, a slope-dependent cost function is applied that results in LCPs including hairpin curves if the slope exceeds a predefined critical value. It is shown that the best-performing LCP sets for the two topographic data sets are closely related but not identical. Moreover, reasons for the failure of LCP reconstructions for some old route sections are discussed.