Hydroclimatology of flow events in the Gila River Basin, Central and Southern Arizona

Traditional flood-frequency techniques are based on the assumption that the observed flood record represents a sample that has been drawn from a single climatically homogeneous population of floods. A hydroclimatic approach was used to evaluate this assumption by identifying the circulation patterns and atmospheric flood-generating mechanisms which control the temporal and spatial variability of flooding. Mean monthly discharges and instantaneous peak flows of the partial duration series were analyzed for thirty gaging stations in the climatically sensitive, semiarid, Gila River basin for the period 1950 to 1980. Correlation fields and composite maps were constructed to define the relationship between 700 mb height circulation anomalies and mean monthly streamflow. Individual flood events were linked to climate by analyzing daily synoptic weather maps and classifying each flood event into one of eight hydroclimatic categories on the basis of the atmospheric mechanisms which generated each flow. The analysis demonstrated that floods and anomalously high streamflow in the Gila River basin originate from a variety of atmospheric processes which vary spatially, seasonally, and from year-to-year. The mechanisms most important for generating floods included winter fronts, cutoff lows, tropical storms, snowmelt, and widespread and localized summer monsoon-related circulation patterns. When flood discharges were grouped into hydroclimatically homogeneous categories, histogram plots of their frequency distributions exhibited means and variances that differed from those of the overall frequency distribution of the entire flood series. The means of the discharges generated by frontal precipitation and tropical storms tended to plot above the mean of the overall series, while the means of floods generated by snowmelt tended to plot below the overall mean. Flood estimates computed from a series containing mixed distributions were not the same as flood estimates computed from climatically homogeneous subsets of the same series. These results have implications for traditional flood-frequency analysis and other stochastic methods of analyzing hydrologic time series. The hydroclimatically-defined subgroups in the flood series of the Gila River basin indicate that nonhomogeneity and nonstationarity can be imparted to a hydrologic time series by differing atmospheric mechanisms alone.