Use of models in large-area forest surveys: comparing model-assisted, model-based and hybrid estimation

• DOI: 10.1186/s40663-016-0064-9
• Publication Date: 2016-02-18T20:39:53Z
• Authors: Göran Ståhl; Svetlana Saarela; Sebastian Schnell; Sören Holm; Johannes Breidenbach; Sean P. Healey; Paul L. Patterson; Steen Magnussen; Erik Næsset; Ronald E. McRoberts; Timothy G. Gregoire
• Keywords: Small Area Estimation

This paper focuses on the use of models for increasing precision estimators in large-area forest surveys. It is motivated by availability remotely sensed data, which facilitates development predicting variables interest We present, review and compare three different estimation frameworks where play a core role: model-assisted, model-based, hybrid estimation. The first two are well known, whereas third has only recently been introduced Hybrid inference mixes design-based model-based inference, since it relies probability sample auxiliary data model target variable from data..We studies surveys based estimation, discuss advantages disadvantages approaches. conclude that no general recommendations can be made about whether or should preferred. choice depends objective survey possibilities to acquire appropriate field data. also modelling approaches successfully applied estimating such as growing stock volume biomass, adequately related commonly available thus purely remain important several parameters.