The use of multiple imputation in the Southern Annual Forest Inventory System
| Authors: | Gregory A. Reams, Joseph M. McCollum |
| Year: | 2000 |
| Type: | Scientific Journal |
| Station: | Southern Research Station |
| Source: | In Hansen, Mark; Burk, Thomas, eds. Integrated tools for natural resources inventories in the 21st century: an international conference on the inventory and monitoring of forested ecosystems; 1998 August 16-19; Boise, ID. Gen. Tech. Rep. NCRS-212. St. Paul, MN: U.S. Department of Agriculture, Forest Service, North Central Research Station: 228-233. |
Abstract
The Southern Research Station is currently implementing an annual forest survey in 7 of the 13 States that it is responsible for surveying. The Southern Annual Forest Inventory System (SAFIS) sampling design is a systematic sample of five interpenetrating grids, whereby an equal number of plots are measured each year. The area-representative and time-series nature of the SAFIS plot design offers increased flexibility in analyzing the data for both large- and small-domain means. Users of FIA information are often interested in the estimates of small-domain means, at the multi-county or FIA survey unit level. Restricting analyses to the most recently measured annual panel results in many missing cells in standard inventory tables. Rather than treat the four unmeasured panels as missing, imputed values are used to update plots in all panels. An initial set of rules and solutions for imputing are provided for SAFIS.