Hedström Ringvall, Anna
- Department of Forest Resource Management, Swedish University of Agricultural Sciences
Ringvall, Anna; Ståhl, Göran; Lämås, Tomas
Guided transect sampling was recently proposed as a method for sampling sparse and geographically scattered populations if auxiliary data with high spatial resolution are available prior to the inventory. In practice, a surveyor will use GPS to follow predetermined survey lines. However, the surveyor will never be able to follow the lines exactly, and thus the accuracy of estimates is likely to be affected. The aim of this article was to study the loss in accuracy of estimates in guided transect sampling due to errors in navigation. The evaluation was made through Monte-Carlo simulations in different simulated forest types. The auxiliary data were considered to be volumes of deciduous trees, and the variable of interest the abundance of an indicator species connected to deciduous trees. To describe the surveyors' movements along the survey lines, a Brownian bridge process was proposed and evaluated with field data. The results from the study indicate that errors in navigation only have a minor effect on both bias and standard errors of estimates of the population total. The bias was not higher than 2.9%, and the increase in standard error not higher than 3.2% in any forest type.
Brownian bridge process; GPS-navigation; multiresource forest inventory; strip surveying; two-stage design
2002, Volume: 48, number: 1, pages: 101-110
Publisher: SOC AMER FORESTERS