Title: Post-Modeling Histogram Matching of Maps Produced Using Regression Trees

Author: Lister, Andrew J.; Lister, Tonya W.

Year: 2006

Publication: In: Proceedings of the sixth annual forest inventory and analysis symposium; 2004 September 21-24; Denver, CO. Gen. Tech. Rep. WO-70. Washington, DC: U.S. Department of Agriculture Forest Service. 126p.

Abstract: Spatial predictive models often use statistical techniques that in some way rely on averaging of values. Estimates from linear modeling are known to be susceptible to truncation of variance when the independent (predictor) variables are measured with error. A straightforward post-processing technique (histogram matching) for attempting to mitigate this effect is presented, and a comparison with untransformed model estimates is made. Histogram matching enhanced the contrast visible in the final map and produced estimates that mimicked the range of estimates in the original data set but performed worse overall with respect to absolute error of prediction. Examples of cases where histogram matching might be an effective post-processing method are given.