Conceptual and numerical models were developed to understand and simulate monthly flow-weighted dissolved-solids concentrations in the Colorado River at Imperial Dam. The ability to simulate dissolved-solids concentrations at this location will help the Bureau of Reclamation satisfy the binational agreement on the volume and salinity of Colorado River water delivered to Mexico. A robust spatial- and temporal-resolution dataset that consists of river discharge and dissolved-solids concentration and load information between January 1990 and September 2016 for 10 sites on canals, drains, tributaries, and the main stem of the Colorado River between Hoover and Imperial Dams was generated. Daily mean dissolved-solids concentrations were estimated and monthly mean dissolved-solids loads were computed for each site. Spatial and temporal load patterns, and historical and current controls on loads and concentrations, were analyzed in order to develop a conceptual model of dissolved-solids transport between Hoover and Imperial Dams. Two numerical models describing the relations between dissolved-solids concentrations and components controlling dissolved-solids concentrations and loads were developed, calibrated, and verified.

Between January 1990 and September 2016, there was a 98.8-million-acre-feet loss of water and a 57.0-million-ton loss of dissolved-solids load from the Colorado River between Hoover and Imperial Dams. Between Hoover and Parker Dams, about 69.0 million acre-feet of water was lost and 51.1 million tons of dissolved solids were lost; between Parker and Imperial Dams, about 29.8 million acre-feet of water was lost and 5.9 million tons of dissolved solids were lost. Water was removed from the river at a relatively consistent rate over the 25-year study period through water transfers to California and Arizona, evapotranspiration from crop irrigation, transpiration processes of riparian vegetation, and evaporation from the river main stem. Dissolved solids were removed from the river between Hoover and Parker Dams at a relatively constant rate through water transfers to California and Arizona, and water pumped from the river for irrigation within the Mohave Valley. A small amount of dissolved solids are gained by the river from inflow from the Bill Williams River. Between Parker and Imperial Dams, however, dissolved solids were not removed from the river at a consistent rate over the study period. Dissolved solids were generally removed from the river from 1990 to 2012, then gained by the river from 2012 to 2015, and then removed from the river from 2015 through 2016. Dissolved solids are assumed to be removed from the river and accumulated within the floodplain sediments and aquifers during irrigation processes; some dissolved solids may also be removed from the river through uptake by crops and riparian vegetation. Dissolved solids accumulated on the landscape and in the floodplain aquifer during irrigation are transported to the river during periods when the hydraulic gradient between the floodplain aquifer and the river is increased, causing a gain in dissolved solids in the river. Dissolved-solids gains in the river occur during periods of relatively low river discharge, such as during the winter months and during drier climatic conditions.

Two numerical models were developed and coefficients were estimated by using data from a May 2008-September 2016 calibration period. One model simulates concentrations at Imperial Dam based on the Colorado River system downstream from Parker Dam, and the other model simulates concentrations at Imperial Dam based on the Colorado River system downstream from Hoover Dam. Both models simulated monthly flow-weighted concentrations of dissolved solids for the Colorado River at Imperial Dam, which corresponded well with observed concentrations for the entire study period. The models are more sensitive to input variables of monthly discharge of the Colorado River below Parker Dam and monthly flow-weighted dissolved-solids concentrations of the Colorado River below Hoover Dam and Parker Dam than to the rate of change in concentration with respect to time and the combined discharge of the Colorado River Indian Reservation Main Canal and the Palo Verde Canal. The calibrated models can be used to run scenarios of future monthly flow-weighted dissolved-solids concentrations in the Colorado River at Imperial Dam. Although the models are expected to provide concentration estimates within 18 milligrams per liter (Parker Dam to Imperial Dam model) to 22 milligrams per liter (Hoover Dam to Imperial Dam model), 95 percent of the time, the error of future scenarios increases as uncertainty in the estimated future input variables increases.