Rocky Mountain Research Station Publications
RMRS Online Publication
RMRS-RP-81: Recursive restriction estimation: an alternative to post-stratification in surveys of land and forest cover
Czaplewski, Raymond L. 2010. Recursive restriction estimation: an alternative to post-stratification in surveys of land and forest cover. Res. Pap. RMRS-RP-81. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 32 p.
Numerous government surveys of natural resources use Post-Stratification to improve statistical efficiency, where strata are defined by full-coverage, remotely sensed data and geopolitical boundaries. Recursive Restriction Estimation, which may be considered a special case of the static Kalman filter, is an attractive alternative. It decomposes a complex estimation problem into simple components that are sequentially processed. Compared to Post-Stratification, it more efficiently uses remotely sensed data, both continuous and categorical. It is less constrained by sample size, which is especially important with panel surveys. It produces a conditionally unbiased covariance matrix for the vector estimate of population totals without approximations or ad hoc assumptions. This facilitates variance estimates for non-linear pseudo-estimators. A robust sequential algorithm controls numerical errors inherent with Recursive Restriction Estimator, which can otherwise cause unreliable results. Analysis of residuals can detect other anomalies.
Keywords: NFI, accuracy assessment, systematic sampling, auxiliary-variables, constraints
About PDFs: For best results, do not open the PDF in your Web browser. Right-click on the PDF link to download the PDF file directly to your computer. Click here for more PDF help or order a printed copy of this publication.
PDF File Size: 460 K
Recursive restriction estimation: an alternative to post-stratification
in surveys of land and forest cover
Electronic Publish Date: September 7, 2010
Last Update: September 7, 2010
RMRS Publications | Order a publication | Contact Us