Reducing the sensitivity of the water quality index to episodic events

Abstract

We demonstrated a general relationship between water quality index (WQI) values and indices of benthic community composition for a set of 32 streams from British Columbia and Ontario. Streams that produced lower WQI values tended to have benthic communities characteristic of degraded water quality. Streams, in contrast, that produced higher WQI scores tended to have a fauna characteristic of high water quality. Trimming the water quality data for high-total suspended solids (TSS) events increased the WQI values by as much as 30 points. There were modest but apparent increases in the strength of the association between the WQI and indices of benthic community composition when the water quality data records were trimmed of values that occurred during periods of high (extreme) turbidity. Trimming data that contained turbidity (or TSS) values beyond the mirrored 5th, 90th, or 95th percentile were about equal in their effect on the WQI. The removal of high TSS samples on the basis of the ‘mirror’ method can be recommended on the basis that it will likely correctly remove data that have a long right-hand tail, and will also correctly not remove data when the data are more normally distributed.