ABSTRACT This paper develops techniques for the mapping of forest types in Arizona, New Mexico, and Wyoming. The methods,involve regression-tree modeling ,using a variety ,of remote ,sensing and GIS layers along with Forest Inventory Analysis (FIA) point data. Regression-tree modeling ,is a fast and efficient technique of estimating variables for large data sets with high accuracy levels. If the methods developed in this paper are successful, they will be applied to the contiguous United States and Alaska producing a forest type map,for these areas. This forest type map will update and improve an older version of the forest type map made in 1992.

ABSTRACT Recent emphasis ,has been placed on estimating ,amount ,and characteristics of forests affected by wildfire ,in the Interior West. Data collected by FIA is intended ,for estimation over large geographic areas and is too sparse to construct sufficiently precise estimates within burn perimeters. This paper illustrates how ,recently built MODIS- based maps of forest/nonforest and biomass coupled with field collected plot data can be used to produce estimates of forest biomass within small geographic areas. FIA-collected forest attributes were modeled,as functions of 250-m resolution digital ancillary variables using nonparametric tree-based methods, then maps were built over diverse ecological mapping,zones. A composite,estimation approach was applied to balance the potential bias of synthetic estimates (generated solely from the maps) against the instability of a ,direct estimator (generated from small numbers,of FIA plots). Methods are described and illustrated through applications to fires that occurred in the years 2002 and 2003.