Geographically Weighted Regression to Explore Spatially Varying Relationships of Recreation Resource Impacts: A Case Study from Adirondack Park, New York, USA

Empirical studies based on spatial explorations have played a critical role in understanding dynamics of recreation resource impact and recovery at multiple scales. However, little research has been done to examine spatially varying relationships between resource conditions and associated geospatial variables, especially using a predictive modeling approach. The primary purpose of this study was to explore spatially varying relationships of recreation resource impacts by using a geographically weighted regression (GWR) model. Specifically, the study was designed to compare the GWR with an ordinary least squares (OLS) multiple linear regression model to better understand localized spatial variations with roadside campsite conditions in Adirondack Park, NY, USA. Geospatial variables contained in the OLS model explained approximately 22% of the variance in campsite conditions (adjusted R2 = 0.220, p < 0.001). Statistically significant predictors of the campsite condition at the global scale included site circumference, distance from water resource, distance from major road, distance from hosting forest road, and slope. Non-significant variables included site designation, distance from recreational trail, and elevation. The subsequent analysis using the GWR model resulted in adjusted R2 values ranging from 0.198 to 0.271 (mean = 0.221). Roadside campsites located in the northern region of the park exhibited relatively higher R2 values, and roadside campsites located in the southern region exhibited relatively lower R2 values. All of the statistically significant global variables showed spatially varying relationships with the campsite condition. Additionally, elevation and site designation factors in the GWR model, which were non-significant variables at the global scale, suggested localized spatial variations with the campsite condition. Overall, the GWR model provided a more robust examination of campsite condition by accounting for localized spatial variations and by improving the model performance. This paper provides a discussion of the methodological and resource management implications of these findings.