Adapting a regional water-quality model for local application: A case study for Tennessee, USA

Abstract

We evaluated whether SPAtially Referenced Regression On Watershed attributes (SPARROW) models calibrated for two adjacent USA regions could be applied at the local scale to support management decisions for streams in Tennessee. Nutrient-source apportionment of load is important for this local-scale application and demands careful consideration of uncertainty in the calibrated coefficients. We used Gauss-Newton regression to test the published SPARROW models for constancy of coefficient estimates between calibration sites on streams within (n = 59) versus outside (n = 327) Tennessee and concluded source apportionment was unbiased for Tennessee streams. The SPARROW models were then applied without re-calibration to predict stream loads and source apportionment for Tennessee streams and to build tools for displaying model results and evaluating source-change scenarios. This approach may inform the adaptation of other regional-scale regression models for use to address water-resource management issues in smaller-scale watersheds.