Impervious Surface Cover: Effects on Stream Salamander Abundance and A New Method of Classification Using Feature Analyst.

Abstract

Increasing impervious surface cover associated with urbanization degrades water quality and alters wildlife habitat. Forested riparian buffers are one popular method of mitigating the effects of impervious surface cover. My study was designed to investigate the effects of impervious surface cover and forested riparian buffer width on the abundance of larval southern two-lined salamanders (Eurycea cirrigera). I sampled 50-meter reaches of 43 streams, representing the range of impervious surface cover and forested riparian buffer width combinations across Wake County, North Carolina, USA in 2004. Additionally, I measured physical and chemical stream properties to account for local habitat effects.

Percent impervious surface cover in the catchment, percent detritus cover in the stream, percent pebble substrate in the stream, and average water conductivity were significant predictors of larval E. cirrigera abundance. Forested riparian buffer width was not a significant predictor. Larval E. cirrigera abundance was lower than anticipated in several streams that appeared to provide good habitat. I discovered that salamander abundance was low in intermittent streams with substrate that is highly sedimented below the surface layer. I suspect that intermittency combined with filled substrate interstices reduced the ability of salamanders to migrate with the water column during dry periods, resulting in low abundances. My research is consistent with a growing body of literature documenting the negative effects of impervious surface on stream biota. In my study, low flow events significantly affected larval E. cirrigera abundance. My findings also suggest that salamander abundance cannot be predicted by measuring the forested buffer width only at the sampling location. A catchment-wide quantification of the stream buffer system, accounting for culverts and other breaches, might yield a better predictive model.

Environmental, economic, and resource management research and decision-making requires current, accurate spatial information. Mapping with aerial imagery can play an important role in providing such information, if it can be done in a cost effective and accurate manner. For my research, I needed an efficient technique for identifying impervious surfaces in urbanizing Wake County, North Carolina, USA. The objectives of my study were to develop an accurate classification of impervious surface using high-resolution aerial imagery and to provide insight into the practical application of Feature Analyst, an object-oriented classification extension within ArcGIS. Feature Analyst utilizes advanced, object-oriented classification algorithms that incorporate spatial context with color and tone to classify high-resolution imagery. I trained Feature Analyst to differentiate pervious and impervious surfaces and classified 111 United States Geological Survey images with a spatial resolution of 33-cm. My classification results yielded an overall accuracy of 92%, with a user's accuracy of 95.2% for the impervious surface class. These results show improvement over historical accuracies of 85% or less. Feature Analyst is most beneficial when spectral value alone will not distinguish classes, object-oriented output is desired, and the time or knowledge for developing complex, customized algorithms is unavailable. I recommend Feature Analyst for classification of high-resolution imagery when object-oriented output is desired and hand-digitizing is impractical.