Regression equations for estimating the 4-day, 3-year low-flow frequency and adjusted harmonic mean streamflow at ungaged sites for unregulated, perennial streams in New Mexico

The Federal Clean Water Act stipulates that States adopt water-quality standards to protect and enhance the quality of water in those States and to protect water quality through the creation of planning documents and discharge permits. Critical low-flow values, including the 4-day, 3-year low-flow frequency (4Q3) and harmonic mean streamflows, are necessary for developing those planning documents and permits. The U.S. Geological Survey computed the 4Q3 and adjusted harmonic mean streamflows using data from 96 streamgages on perennial streams, and regression equations were developed for the estimation of these parameters at ungaged, perennial streams in the State of New Mexico using weighted least-squares regression and readily accessed basin and climatic characteristics. Six equations were developed for the 4Q3 statistic, and five equations were developed for the adjusted harmonic mean statistic. Separate equations were developed for sites located in basins with mean elevations equal to or greater than 8,000 feet above the National Geodetic Vertical Datum of 1929 (except where noted as the North American Vertical Datum of 1988), as well as for sites on streams that are tributary to the San Juan River. Pseudo R-squared values ranged from 0.53 to 0.87 (4Q3) and adjusted R-squared values ranged from 0.69 to 0.89 (adjusted harmonic mean). For sites in basins with mean elevations of less than 8,000 feet above the National Geodetic Vertical Datum of 1929 (except where noted as the North American Vertical Datum of 1988), equations were developed based on contributing drainage area size. Drainage area, mean basin elevation, basinwide mean annual precipitation, and mean basin slope were found to have relations to the 4Q3; drainage area, mean basin elevation, basinwide mean annual precipitation, mean basin slope, and mean basinwide precipitation for the winter period, defined as the months of October through April, were found to have relations to the adjusted harmonic mean. Comparison to previous 4Q3 regression equations using fit statistics indicate an overall improvement in performance.