You are here:
Quantifying Urban Watershed Stressor Gradients and Evaluating How Different Land Cover Datasets Affect Stream Management
Citation:
Smucker, N., A. Kuhn, M. Charpentier, C. Cruz-Quinones, Colleen M. Elonen, S. Whorley, Terri M. Jicha, Jonathan R. Serbst, Brian H. Hill, J. Wehr, AND Jim Lake. Quantifying Urban Watershed Stressor Gradients and Evaluating How Different Land Cover Datasets Affect Stream Management. ENVIRONMENTAL MANAGEMENT. Springer-Verlag, New York, NY, 57(3):683-695, (2016).
Impact/Purpose:
Watershed development and associated activities are leading sources of negative impacts to stream ecosystems, making effective monitoring and water quality models necessities to making informed decisions. Here, we developed a sampling scheme that adequately captured disturbance gradients in an efficient manner, which can benefit state and other monitoring agencies seeking to develop water quality criteria and ecological indicators, while helping to optimize the use of limited funds and personnel. We also identified important differences in land cover estimates and predictive water quality models that resulted from two data sources with different land cover resolution (30 meter versus sub-5 meter resolution). Misestimated land cover could affect resource protection due to misguided management targets, watershed development practices, or water quality criteria.
Description:
We used a gradient (divided into impervious cover categories), spatially-balanced, random design (1) to sample streams along an impervious cover gradient in a large coastal watershed, (2) to characterize relationships between water chemistry and land cover, and (3) to document differences between estimates of watershed land cover using 30-m resolution National Land Cover Database (NLCD) and ≤ 5-m resolution land cover data and how differences affect predictive models of stressors. Increased concentrations of nutrients, anions, and cations were significantly correlated with increased watershed percent impervious cover (IC), regardless of data resolution. The NLCD underestimated percent forest for 71/76 sites by a mean of 11% and overestimated percent wetlands for 71/76 sites by a mean of 8%. The NLCD almost always underestimated IC at low development intensities and overestimated IC at high development intensities. As a result of underestimated IC, regression models using NLCD data predicted mean background concentrations of NO3– and Cl– that were 475% and 177%, respectively, of those predicted when using finer resolution land cover data. Our approach to sampling design could help states and other agencies seeking to create monitoring programs and indicators responsive to anthropogenic impacts. Misestimated land cover could affect resource protection due to misguided management targets, watershed development and conservation practices, or water quality criteria.
