SOME ASPECTS OF LINE INTERCEPT SAMPLING

Many procedures exist for sampling vegetation. Most of them may be categorized into one of five categories: quadrat, point, distance, line transect, and line intercept sampling methods. The line intercept method is emphasized in this thesis.;The results in this thesis are contained in chapters II-IV. Each chapter is designed to be a paper in itself. Chapter I is an introductory chapter and chapter V summarizes the contents of the thesis.;Experimentors have been using cost efficient systematically located transects for the line intercept method for some time with little support from mathematical statistics. In chapter II it is shown that for rectangular study regions the usual line intercept estimators for density, percent cover, and other attributes are unbiased for a sample of systematically located transects and are approximately unbiased for most "large" irregularly shaped study regions.;Even though the usual estimators are unbiased or nearly so, systematic sampling may introduce dependencies into the transect data. It is possible that the estimates for density, percent cover, or another attribute on one transect are correlated with the estimates on a nearby transect. This phenomenon of correlation in space is attributed to regionalized variables and random functions are used to model the deterministic and random components of the regionalized variables. Associated with the theory of regionalized variables is the estimation technique developed by the South African statistician and mining engineer D. G. Krige. Chapter III introduces the semi-variogram and kriging as a means to utilize the correlation which may exist between transects.;Line intercept sampling has been primarily used in the estimation of density, percent cover, and any other fixed measurable attribute. The unbiased estimators commonly used are developed under the assumption of a randomly located transect or a sample of randomly located transects. Variance formulas have also been developed under the assumption of randomly located transects and random (uniform) distribution of particles.;Chapter IV addresses three interrelated problems. First, an approximate Chi- Square test to test for the randomness of the particles is given. The hypothesis of randomness will be rejected for large values of the statistic if the spatial distribution is contagious. If the spatial distribution follows a fixed regular pattern, the hypothesis of randomness will be rejected for small values of the test statistic.;Second, an unbiased estimator for the variance of the estimator for other attributes is derived. And third, two methods for estimating the variances of the estimators for density, percent cover, or another attribute are introduced which do not depend on random (uniform) distribution of particles. The first of these techniques is based on the theory of chapter III and the estimated semi-variogram. The second procedure is more suitable for estimating the variance of the cover estimator; however, it may also be used for the other two estimators. This procedure is based on treating the measured values as sequences of values and using the finite fourier transform.