Factors Influencing the Precision of Water Quality Assessments and Benchmarks

Ecological assessments generally seek to characterize reference conditions and then determine condition by assessing departure from those reference conditions. However, conditions at reference sites are not static through time and can be influenced by diffuse anthropogenic factors such as climate change. Ideally managers would be able to determine impacts of climate change and direct management actions on ecological condition separately. This would ideally occur by using reference data collected as far back in time as possible to reduce impacts of climate change on sites considered to otherwise be in reference condition. However, all index development requires a trade-offs between getting adequate sample sizes of reference sites and maximizing reference quality.

For this project, we sought to revise water quality models in Olson and Hawkins 2012 and 2013 to be based off static 30 year windows of climate predictors instead of climate data from the year of data collection which could bias predictions of natural conditions by accounting for anthropogenic changes in climate. Additionally, we conducted a sensitivity analysis to maximize reference site representativeness and model performance while only using the oldest data possible. We compare model predictions for these new models to the old models of similar performance to determine how model revisions impact benchmarks that the BLM uses for water quality assessments. Furthermore, we compare four ways of calculating prediction intervals for models to determine appropriate water quality benchmarks for use by state and federal agencies such as the BLM.