Ecologically-focused calibration of hydrological models for environmental flow applications

Includes bibliographical references.2015 Fall.Hydrologic alteration resulting from watershed urbanization is a common cause of aquatic ecosystem degradation. Developing environmental flow criteria for managing the effects of urbanization and other human influences requires quantitative flow-ecology relationships that link biological responses to streamflow alteration. To the extent possible, gaged flow data are used; however, bioassessment sites are frequently ungaged and hydrological models must be used to characterize flow alteration. Physically-based rainfall-runoff models typically utilize a "best overall fit" calibration criterion, such as the Nash-Sutcliffe Efficiency (NSE), that does not focus on specific aspects of the flow regime relevant to biotic endpoints. This study aims to identify how accurately coastal southern California rainfall-runoff models can be calibrated using specific elements of the flow regime known a priori to be critical to benthic macroinvertebrates (ecologically-focused) versus a traditional best overall fit criterion. Additionally, this study seeks to assess the utility of ecologically-focused calibrated models by comparing flow metric accuracy and the strength of flow-ecology relationships among different calibration approaches versus gage data. For this study, continuous HEC-HMS 4.0 models were created for 19 coastal southern California watersheds and calibrated to USGS streamflow gages with nearby bioassessment sites using one best overall fit and three ecologically-focused criteria: NSE, Richards-Baker Flashiness Index (RBI), percent of time when the flow is < 28 L/s (< 1 cfs), and a Combined Calibration (RBI and < 1 cfs), respectively. Ecologically-focused criteria were selected based on preliminary statistical flow-ecology relationships at gaged bioassessment sites. Calibrated models were compared using flow metric accuracy relative to gage data and the strength of flow-ecology relationships. Models were highly accurately calibrated to ecologically-focused criteria, with calibration median percent errors less than 1.5% and only a single model with a percent error greater than 10%, and NSE criteria, with a median value of 0.634. Regardless of high calibration accuracy for ecologically-focused models, additional flow metrics not explicitly calibrated, especially those describing magnitude or rise and fall rates at aggregated daily time scales, were not consistently reproduced by models. Despite inaccuracies across a full suite of 71 flow metrics, low flow and flashiness metrics relevant to biotic endpoints were modeled accurately (< 20% error) and often provided stronger flow-ecology relationships than best overall fit criteria in terms of adjusted R2 in multiple regression analyses and variance explained in random forest modeling. This was especially true when two ecologically-focused criteria were combined, suggesting the importance of multiple calibration criteria. Flow metrics from the Combined Calibration provided the strongest flow-ecology models in correlation and regression analyses compared to the other three calibration approaches, and perform similarly in random forest models. This study demonstrates that if ecologically relevant flow metrics can be identified using published literature or preliminary statistical analyses of gaged bioassessment sites prior to developing a hydrologic foundation, they can be incorporated as calibration criteria and provide stronger modeled flow-ecology relationships than exclusive use of a best overall fit criterion