Truncation Effects on Estimated Parameters of Tracer Distributions Sampled on Finite Domains

Abstract

Methods are presented and demonstrated for compensating for the apparent effect of truncation of the sampled crosswind distribution of a tracer, at either or both boundaries of a surface sampling grid downwind from the point of release. Errors and correction on estimates of the apparent vertical mixing depth, h, the location of the centroid of the crosswind distribution ?, and the crosswind dispersion parameter, σy are those which would apply to a truncated Gaussian distribution having the observed moments (zeroth, first and second). For truncation on only one side of the tracer cloud, a linear relationship between certain measurable quantities based on moments of partial distributions is used to derive the parameters of a best-fit truncated Gaussian distribution by linear regression. The methods are illustrated by an example from the Cross Appalachian Tracer Experiment and by a Gaussian simulation.