Horizontal Line Sampling for Riparian Forests without Land Area Estimation

Abstract

Abstract Sampling forests in riparian areas presents unique difficulties that can be addressed by judicious application of line sampling. Land areas associated with riparian forests often have highly irregular boundaries, making accurate area determination difficult. Therefore, estimators for total amounts of forest attributes that do not depend on land area are desirable. Such estimators can be obtained by using horizontal line sampling from sample lines that are located at random intervals along a straight baseline and which extend entirely through the riparian area of interest. Horizontal line sampling selects sample trees that subtend an angle whose vertex is located on a length of line. This is conceptually similar to Ståhl's transect relascope sampling (Ståhl, G. 1998. Transect relascope sampling—A method for the quantification of course woody debris. For. Sci. 44:58–63), which estimates total amounts of coarse woody debris from lines established at fixed intervals. Use of a control variate or importance sampling based on predicted line length (where available) is suggested as a variance reduction technique. An unbiased estimator for sampling at intervals along the line is suggested. Monte Carlo integration estimates of riparian land area are proposed for estimation of per-unit area quantities. These estimators could be applied to any forested area but appear to be most well suited to long, relatively narrow riparian areas.