Extensive overstory mortality of quaking aspen (Populus tremuloides Michx.) surrounding southern Utah’s Cedar Mountain has incited concern among stake holders. This decline has proliferated during the opening decade of the 21st Century and may be correlated with periodic drought. The present study attempts to explain temporal aspects of the aspen decline through a remote sensing based change detection analysis. The main objectives were to determine the proportion of aspen contained in a determined study area of Cedar Mountain in 1985 and to model how the proportion has changed over the past 23 years. To accomplish these objectives, 46 in situ verified aspen points and 68 non-aspen points were “drilled” through reflectance data (from Landsat TM 1985-2008) for the year in question along with DEM derivatives. Subsequently, data mining software was used to create classification trees, based on the files produced through drilling. Thereafter, binary aspen presence maps for 1985, 1990, 1995, 2001, 2003, 2005, and 2008 were created. Additionally, an aspen habitat model was constructed using reflectance data and topographic layers to constrain analysis exclusively to areas of aspen cover. Change detection was achieved by combining two images of the time series, using a post-classification comparison technique.
Results/Conclusions
The 1985 model contained the highest proportion of aspen cover (0.948 or 10,910.3 hectares), while the 2008 model had the lowest proportion (0.725 or 8,339.2 hectares). The cumulative decrease in aspen cover over the time series was 23.6% and the mean annual decrease was 1.02%. Aspen cover decreased in a step-wise fashion over the duration of the time series. The post-classification comparison showed dramatic losses in aspen cover between the periods of 1985 to 1990, 1990 to 1995, and 2001 to 2003. Conversely, the periods of 1995 to 2001 and 2003 to 2008 had negligible changes in aspen cover. The period with the greatest decline was the 2001 to 2003 period, with a net loss of 867.96 hectares. While some of the differences observed in the models may be attributed to year-to-year differences in precipitation or canopy growth, the unquestionable negative trend in aspen cover portrayed in the models substantiates stake holders’ assertions of aspen mortality. Preliminary analysis of precipitation data suggests that the step-wise pattern of aspen decline on Cedar Mountain may be correlated with the timing in drought cycles.
