Implications of climate variability for low flows and dilution of discharges from point sources

The National Pollution Discharge Elimination System (NPDES) was created through the Clean Water Act as a mechanism for regulating the release of toxic substances to streams in the U.S. The concentration of each toxic substance in a wastewater discharge is governed by an NPDES permit, which is intended for the protection of designated uses such as maintenance of aquatic life. The allowable concentration limit in the effluent is usually determined with a steady-state model using mass balance calculations, one element of which is the dilution flow available in the receiving water. Dilution flow is defined by analysis of historical data using a specific set of critical, low-flow conditions. The use of historical data implies that the past is a good indicator of future conditions, at least for the duration of a discharge permit. In most cases, the hydrologic record that can be used for the analysis extends no more than a decade or two, depending on the availability of gage records and recent alterations to the flow regime (e.g., dams, diversions, etc.). Short records introduce great uncertainty in the estimation of low flows because they are unlikely to capture events with periodicities of multiple years. For example, ENSO events, which have a frequency of 4-6 years, would be poorly represented by hydrological records that did not extend several decades. Major droughts are even less frequent.

We conducted an analysis of daily flows at several gages with long records in the South Platte basin of Colorado. Low flows were calculated for successive 10-y blocks of data, and these were compared with low flows calculated for the entire period of record (>70 years). Historical variability of stream flows is great enough that it raises concerns about the extent to which low flows derived from a 10-y record will be protective of water quality. In unregulated streams, there was no trend over time, but the low flows for a single 10-y block could differ from the long term value by as much as a factor of 2. Low flows taken from a 10-y block selected at random would be protective of aquatic life in fewer than half of the years for the full period of record. The hydrographs of most streams in Colorado have been influenced by dams, diversions, or water transfers. These alterations to the natural flow regime shorten the record available for analysis and generally result in a trend of increasing low flows that is visible when successive 10-y blocks of data are analyzed. Use of a shorter record decreases the chance that significant climate variation will be reflected in the low flows. The presence of a trend based on present patterns of water use may carry an unanticipated risk by failing to incorporate societal response to severe drought conditions, for example. There is a clear need for a mechanism that will incorporate the effects of climate variation in the determination of low flows for NPDES permits.