Targeting During Sampling and Analysis in Fire History Reconstructions and Fire-Climate Research: Evaluating Methodologies.

Fire scar datasets have been widely applied to the reconstruction of historical fire regimes, in the investigation of fire-climate linkages and in environmental assessments. These datasets have been collected by numerous methods, yet the spatial and statistical integrity of these methods have not been evaluated. Targeting, an experience-based sampling system, has been explicitly recommended in the literature. Logistically, targeting is a practical method in study areas that are often characterized by difficult terrain. However, targeting may bias datasets by emphasizing areas of the landscape that burn more frequently. Uniform grid-based sampling methods have also been implemented in fire research, and may better represent fire heterogeneity. In this paper, I simulate grid-based and targeted sample designs at multiple scales in a GIS. An existing dataset collected in a Pinus ponderosa forest in Washington with 0.073 samples/ha, is treated as the population of scarred trees. Mean fire return interval (MFRI) is calculated for the sampled subsets of data at multiple scales and compared with the MFRI for the entire dataset. The purpose of this study is to evaluate the validity of targeting, and to investigate the relationship between scale and sampling methods. Results suggest that targeted sampling methods do not bias fire intervals.