Systematic Sampling of Scanning Lidar Swaths

Abstract

Proof of concept lidar research has, to date, examined wall-to-wall models of forest ecosystems. While these studies have been important for verifying lidars efficacy for forest surveys, complete coverage is likely not the most cost effective means of using lidar as auxiliary data for operational surveys; sampling of some sort being the better alternative. This study examines the effectiveness of sampling with high point-density scanning lidar data and shows that systematic sampling is a better alternative to simple random sampling. It examines the bias and mean squared error of various estimators, and concludes that a linear-trend-based and especially an autocorrelation-assisted variance estimator perform better than the commonly used simple random sampling based-estimator when sampling is systematic.