Spatial Distribution of Hydrologic Ecosystem Service Estimates: Comparing Two Models
P James Dennedy-Frank, Northeastern University, Marine and Environmental Sciences, Coastal Sustainability Institute, Boston, MA, United States, Yonas Ghile, Stanford University, Los Altos Hills, CA, United States, Steven M Gorelick, Stanford University, Stanford Woods Institute for the Environment, Stanford, United States, Rebecca Logsdon Muenich, University of Arkansas, Fayetteville, AR, United States, Indrajeet Chaubey, Purdue University, Earth, Atmospheric, and Planetary Sciences, West Lafayette, IN, United States and Guy Ziv, University of Leeds, School of Geography, Leeds, United Kingdom
Abstract:
We compare estimates of the spatial distribution of water quantity provided (annual water yield) from two ecohydrologic models: the widely-used Soil and Water Assessment Tool (SWAT) and the much simpler water models from the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) toolbox. These two models differ significantly in terms of complexity, timescale of operation, effort, and data required for calibration, and so are often used in different management contexts. We compare two study sites in the US: the Wildcat Creek Watershed (2083 km
2) in Indiana, a largely agricultural watershed in a cold aseasonal climate, and the Upper Upatoi Creek Watershed (876 km
2) in Georgia, a mostly forested watershed in a temperate aseasonal climate. We evaluate (1) quantitative estimates of water yield to explore how well each model represents this process, and (2) ranked estimates of water yield to indicate how useful the models are for management purposes where other social and financial factors may play significant roles.
The SWAT and InVEST models provide very similar estimates of the water yield of individual subbasins in the Wildcat Creek Watershed (Pearson r = 0.92, slope = 0.89), and a similar ranking of the relative water yield of those subbasins (Spearman r = 0.86). However, the two models provide relatively different estimates of the water yield of individual subbasins in the Upper Upatoi Watershed (Pearson r = 0.25, slope = 0.14), and very different ranking of the relative water yield of those subbasins (Spearman r = -0.10). The Upper Upatoi watershed has a significant baseflow contribution due to its sandy, well-drained soils. InVEST’s simple seasonality terms, which assume no change in storage over the time of the model run, may not accurately estimate water yield processes when baseflow provides such a strong contribution. Our results suggest that InVEST users take care in situations where storage changes are significant.
