Three methods of estimating the current state of forest attributes over small areas for the USDA Forest Service Southern Research Station's annual forest sampling design are compared. The three methods were (I) simple moving average, (II) single imputation of plot data that had been updated by externally developed models, and (III) local application of a global model that was determined through mixed estimation. In a preliminary analysis of current basal area estimation, the less complex Method III compared favorably with Method II in terms of squared error loss.
Roesch, Francis A., Jr. 1999. Mixed Estimation for a Forest Survey Sample Design. Reprinted from the 1999 Proceedings of the Section on Statistics and the Environment of the American Statistical Association