Due to their complex and inhomogeneous characteristics, the propagation and radiation parameters of trees and vegetation areas are very difficult and time consuming to obtain. This communication proposes a statistical method, using robust weighted local regression, to minimize the influence of the effect of the tree inhomogeneity on its re-radiation pattern, allowing the evaluation of averaged re-radiation functions from simple measurements. The proposed method was successfully applied to six tree specimens of both conifer and ficus species, at 60 GHz. Furthermore, once this empirical averaged function is obtained, first-order statistics can be applied to generate several simple reradiation functions, statistically identical to those obtained from measurements. Thus, allowing the full characterization of the tree under study.