Abstract
The health of freshwater systems has been in decline and human activities have been known to be the cause of the degradation of freshwater quality. To maintain our freshwater resources, watershed health assessment efforts are needed. To retain consistency and comparability between watersheds health assessments, construction of a Watershed Health Index (WHI) is an effective method to assess and compare watershed health. Several state agencies around the U.S. have developed WHIs to assess watershed health within their state. In this study, we aimed to develop a WHI for Washington State with past watershed data through data-based approaches. We used the metrics B-IBI as response variables and the watershed health indicators as the predictor variables in Partial Least Square (PLS) regression and Boosted Regression Tree (BRT) to generate relative influences of each indicator on the variability of B-IBI, and compiled the relative influences into weight factors to calculate the final WHI. The results showed low R2 (between 0.05 to 0.30) for the PLS regressions, and the two methods (BRT and PLS) resulted in different weight factors in the WHI. We suspect that the data structure and sample size may have contributed to the low R2, since most samples were non-continuous measurements of watershed indicators, or “snapshots”, which may capture extreme events that may not represent the actual health condition of the watersheds.