Quantification and prediction of lateral channel adjustments downstream from Cochiti Dam, Rio Grande, NM

Abstract

Located downstream from Cochiti Dam in north-central New Mexico, the Cochiti reach of the middle Rio Grande spans 45 kilometers from the dam to the Highway 44 bridge in Bernalillo, NM. During the early and mid-1900's the U.S. Army Corps of Engineers and the U.S. Bureau of Reclamation constructed dams and performed channelization in an attempt to reverse the historic aggradational trend of the channel bed and to minimize flood hazards on the middle Rio Grande. Cochiti Dam, completed in 1973, was designed for sediment detention and flood control and traps virtually all of the sediment entering the Cochiti reach. Few large alluvial rivers in the southwestern U.S. have been studied and documented as well as this section of the Rio Grande. Hydraulic, topographic, photogrammetric and sediment data collection efforts conducted by numerous federal and state agencies have tracked changes in the river since 1895. Utilizing this extensive database, the Cochiti reach of the middle Rio Grande was thoroughly characterized through analysis of flow regime, sediment transport, cross-sectional form, bed material, longitudinal profile, planform, and lateral movement rates. The response of the river to altered hydrologic regimes was studied through application of hydraulic geometry equations, sediment budget analysis, and examination of lateral migration rates. The rates of lateral movement and level of lateral stability of the river were quantified using digitized aerial photos of the active channel over 74 years (1918-1992). Four indices of lateral movement and two indicators of lateral stability were measured from the digitized active channel delineation. Results of the historic geomorphic analysis and the quantification of lateral mobility and stability show that the channel moved toward a more stable state as the peak discharges decreased prior to and following construction of the dam. During the entire time period studied (pre and post dam), the channel narrowed and moved toward an equilibrium width as predicted by hydraulic geometry equations. Lateral movement rates have declined since 1918 and the channel has shifted from a multi-thread to a more single-thread pattern. The lateral changes that began prior to construction of Cochiti Dam appear to be the result of changes in the hydrologic regime rather than the sediment regime. The dam withholds 99% of the sediment entering the reach. Since construction of the dam, the bed has coarsened from sand to gravel size sediment and degraded up to 2 meters. The sinuosity also increased up to 9% between 1972 and 1992. Three models were developed to estimate the lateral migration and width change rates of the Cochiti reach from 1918 to 1985. The 1985 to 1992 data were reserved for validation of the models. The first model was a simple estimation of the movement rates as a percentage of channel width and produced rough estimates of migration and width change rates. The second model was based on the assumption that as the river moved closer to an equilibrium state, the rate of change decrease. Fitting an exponential equation based on deviation from equilibrium produced r-square values as high as 0.98 in modeling the 1918-1985 width changes of the Cochiti reach. Similar models of migration rate produced r-square values up to 0.89 for the same time period. The third model utilized multiple regression analysis with stepwise selection techniques to identify significant associations between lateral movement rates and measures of flow energy, sediment supply, bed material and planform. The resulting equations showed that flow energy (water discharge and/or channel slope) and planform explained the highest percentage (R2 = 0.65) of variance in migration rate. Width change rates were best described by an equation including unit stream power (S(Q)0.5) combined with active channel width (R2 = 0.50). More important than the predictive capabilities of this model were significant associations between lateral movement rates and flow and planform parameters. Migration rates increased with increasing flow energy, sinuosity and total channel width (width of channel including vegetated bars). Width change rate was significantly associated with active channel width only. Validation of the models with the 1985 to 1992 Cochiti reach data showed that the regression results best estimated the migration rate. The equations combining discharge and width ratio or just mobility index produced the smallest average error (25% to 38%). The 1992 active channel width was estimated equally well (12% average error) by the simple model of width change rate as percentage of width and the regression equation using unit stream power and active channel width.