Improvements in the Colorado ammonia model by simultaneous computations of extremes in flow and water chemistry

Abstract

The present study reports the results of an analysis of association for extremes in pH, temperature, and low flow at 12 sites in Colorado for which an extensive data record is available (21 years at 12 or more samples per year). Statistical study of data for these sites showed no general relationship between flow and percent unionized ammonia at any station. Within periods of low flow, there was no parametric association between percent unionized ammonia and low flow. A nonparametric test of association between percent unionized ammonia and low flows that 8 of the stations have a random association of the two variables, i.e., the expected value of percent unionized ammonia during a period of low flow is equal to the mean value rather than an extreme value. At four of the stations, the association of low flow with percent unionized ammonia was nonrandom. Three of these showed a negative association, i.e., the percent unionized ammonia was significantly lower than the mean during periods of low flow. At a single station (the South Platte River near Kersey), there was a significant positive association, albeit a rather loose one, between percent unionized ammonia and extreme low flows; the lowest 5th percentile of low-flow values showed a mean of the 77th percentile unionized ammonia. These findings suggest modifications of models that are used in computing maximum total ammonia for permits. The assumption of strong association between the least favorable flows (low flow) and least favorable percent unionized ammonia (high percent unionized ammonia) is not justified by field information and may result in overly stringent ammonia control requirements for point source discharges.