Advancements in Water Quality Modeling using SPARROW (SPAtially Referenced Regression On Watershed attributes)

The SPARROW water-quality model was originally developed to gain understanding of the sources, transport, and predominance of contaminants in unmonitored streams at large regional or national spatial scales under mean hydrologic conditions. Initial models were developed for total nitrogen and phosphorus using the Reach File-1 stream network. The list of contaminants has since expanded to include salinity, suspended sediment, streamflow, and total organic carbon, and the reach network now commonly in use is the 100K National Hydrography Dataset to include greater than an order-of-magnitude more reach segments and waterbodies. Recent advances in methodology have significantly enhanced modeling capabilities. Model specifications can now be dynamic at seasonal or monthly time steps to allow for the estimation of lags in the release of contaminants from catchment storage that are driven by time-varying climatic and land use conditions. Because computational constraints generally limit the spatial extent that dynamic models can be applied, an additional model option is a mean seasonal or monthly specification that allows the assessment of legacy contaminants across large spatial scales such as hydrologic regions or the nation. The updated SPARROW model can be used to estimate loads for multiple contaminants simultaneously, making it possible, for example, to specify instream organic nitrogen as an additional source for nitrate that augments standard point and non-point sources, or for a single model to produce estimates of total and baseflow loads. The estimation methodology has also been upgraded to allow estimation in the R programming language using Bayesian methods. The Bayesian methods incorporate hierarchical and state-space model specifications that can be used to account for regional variation in model coefficients and uncertainty in instream load estimates that might cause estimation bias. All the enhancements to the original model retain the unique features of SPARROW, using an empirical, statistical mass balance methodology to obtain a more complete understanding of the sources of contaminants and the processes that control their transport through the hydrologic system.