Resource information
The Forest Service, U.S. Department of Agriculture's (USDA's) Northern Research Station Forest Inventory and Analysis program (NRS-FIA) uses digital land cover products derived from remotely sensed imagery, such as the National Land Cover Dataset (NLCD), for the purpose of variance reduction via postsampling stratification. The update cycle of the NLCD product is infrequent; NLCD 2001 was the first update since the release of NLCD 1992, and was not yet fully completed as of late 2006. Consequently, FIA field data collected as recently as 2005 are being poststratified with land cover data collected more than a decade before. In addition, NRS-FIA has performed its own land cover classification of remotely sensed imagery for use in the stratification of some States, a time-consuming process. Alternative sources of information need to be evaluated both to eliminate the temporal mismatch between land cover data and FIA plot information and to reduce the amount of analyst time required to perform the stratification process. The USDA's National Agricultural Statistics Service, in conjunction with the Foreign Agricultural Service, produces the Cropland Data Layer (CDL) using satellite imagery. The product, updated yearly for multiple States, includes detailed classification of crop type and also a "Woodland" category. In this study, the CDL was compared to the NLCD 1992 data set for Wisconsin. This comparison included two components: (1) county-level, pixel-derived estimates of forest land area relative to each other and to plot-based FIA estimates and (2) variance reduction produced by each layer when used in postsampling stratification of FIA plots for estimating primary FIA attributes (forest area, number of trees, timber volume, and tree biomass). Results indicate poor agreement between CDL pixel-based area estimates and FIA plot-based estimates; however, when used for poststratification, the CDL produces similar estimates of primary FIA attributes to those of the NLCD at the State level and higher relative efficiency at the FIA survey-unit level.
