Weisberg, P. J., Ko, D., Py, C. & Bauer, J. M. (2008). Modeling fire and landform influences on the distribution of old-growth pinyon-juniper woodland. Landscape Ecology, 23, 931-943.

 

Weisberg et al.’s study (2008) examined the influences of fire and topography on the distribution of old-growth pinyon-juniper woodland in the western United States using multiple GIS models and statistical models. The objectives were determining which fire risk component (fuels or topography) has been most strongly associated with old-growth distribution, assessing the strength of association between spatial patterns of fire risk and old-growth P-J woodland distribution, as well as comparing the utility of different spatial modeling methods of old-growth distribution. Weisberg et al. maintain that age structure is an important factor in determining the fire frequency, so they hypothesized that “areas of old-growth P-J woodland are restricted to fire-safe areas” (p.932). 

 

The study area was located in the Barrett Canyon watershed in the Shoshone Mountain Range in central Nevada, USA. Weisberg et al. derived a canopy coverage polygon map from photo-interpretation of panchromatic digital orthophoto quadrangles at 1 m x 1 m resolution to delineate age classes of P-J woodland. Age class was divided into young (<150 years), mixed, old (>150 years). Authors further used Fragstats to calculate metrics for comparing patch density, size, and shape among the different age class categories. Three GIS models, a logistic regression model, and a cellular automata model were developed to predict old-growth’s fire susceptibility, but we focused on the review of GIS models.

 

Three GIS models were developed by Weisberg et al. to hypothesize fire risk. Firstly, the fuel loading model calculated topographic convergence index (TCI) and solar radiation index to represent effects of spatial variability in fuel loading on old-growth productivity. High TCI means sites collect and retain water during redistribution events; whereas, low TCI indicates sites with steep slopes. For example, sites on steep slopes were considered to have a low probability of burning due to low fuel loads, and thus to have a high likelihood of old growth trees.

 

The second model was the topographic barriers model, which dealt with the following spatial variables: locations of ridge lines, proximity to rock outcrops, and a wind exposure index. Weisberg et al. used a hydrological modeling routine in ArcGIS to draw ridge lines from a DEM at 30-m resolution. Ridge proximity was a Boolean variable indicating whether a site lines on a ridgeline or not. Rock outcrops proximity indicated whether a site lies within 30 m of a rock outcrop. A wind exposure index described whether the wind is protected or exposed. For instance, fire risks are low if sites were within 30-m of rock outcrops and thus, there is a high old-growth probability. The third GIS model combined the effects of fuel loading and topographic barriers to fire spread.

 

Weisberg et al. also used Akaike’s Information Criterion statistic (AIC) model to predict old-growth probability as a function of topographic variables were compared. It considered all possible combinations of the topographically-derived variables, such as solar radiation (RAD), wind exposure (WIND), topographic convergence index (TCI), proximity to ridge lines (RIDGE), and proximity to large rock outcrops (ROCKS). The study compared these models to each other and to a null model including only an intercept term, as well as to a cellular automata model.

 

As a result, the study concluded that “the spatial distribution of fuels played the stronger role for influencing landscape-level age structure of [P-J] woodland” (940). We would rate this paper 7 out of 10.We thought the authors did an exemplary job not only in using GIS models to present the results but also comparing different models to approach an accurate prediction. Although the study was conducted at a local scale, the models adopted in this study could also be applicable to future analysis in varying subjects. In this study, GIS analysis plays a crucial role in integrating abiotic factors with the spatial content, in which assessing the relationship between fire risks and topographic changes and the distribution of old-growth woodland. Nevertheless, one suggestion that we would like to provide to improve this study is to further run an explanatory regression to explicitly determine which combinations of variables have the most significant effects on old-growth probability.