UBC Theses and Dissertations
An assessment of variable radius plot sampling techniques for measuring change over time : a simulation study Carter, David Hugh Harrison
The most commonly used approach for measuring change over time involves using Fixed Radius Plots (FRPs). A major disadvantage of using FRPs for detecting change is the cost of implementation, both statistically and monetarily. Variable Radius Plots (VRP) can also used for measuring change over time; however, VRPs have not been widely used due to perceived problems including: 1. sudden additions to the value of interest which produce high variability; 2. complex mathematical formulae used for computation; 3. use of an angle for measuring trees; and 4. lack of studies on the use of VRPs for measuring change over time. In this thesis four VRP change estimators were evaluated: (1) the traditional subtraction estimator; (2) the Grosenbaugh estimator; (3) the Distance Variable (DV) estimator; and (4) the Flewelling estimator. Tree and stand data were generated using a stand simulator (StandSim), and stands were sampled using a series of SAS programs. Volume change, basal area change, and stems per hectare change were calculated for a series of 54 stand conditions in which density, dbh distribution, mortality, and spatial distribution were varied. Relative efficiencies were calculated comparing each VRP estimator to both the FRP estimator, and the subtraction estimator. The DV estimator had a relative efficiency greater than 1.0 (i.e., it was more precise) in 59% of the scenarios for volume change, 15% of the scenarios for basal area change, and 0% of scenarios for stems per hectare change when compared to FRP. Mortality is a high source of variability for all estimators and in another comparison where high mortality scenarios were excluded, the DV estimator had a relative efficiency greater than 1.0 in 74% of scenarios for volume change, 30% of scenarios for basal area change, and 0% of scenarios for stems per hectare change. The DV estimator had a relative efficiency greater than 1.0 in at least 70% of the scenarios for each attribute when compared to the traditional subtraction method, excluding the high mortality scenarios. This thesis demonstrates that VRP estimators for measuring change over time (in particular the DV estimator) can reduce the variability introduced by sudden additions of value. VRP estimators are not so complex that they should be excluded as an option for use, and require no new field techniques. Further studies should be completed to support the use of VRPs for change detection.
Item Citations and Data