Comparison of infiltration rates in the dry riverbed of the Colorado River Delta during environmental flows

Water has been over-allocated all over the world; natural rivers have been over-exploited and regulated to manage storage and distribute the water to thirsty cities and agricultural lands. The Colorado River is no exception to this trend. It is one of the most regulated rivers in the world; the once mighty Delta that formed from the river has become a few hectares of native trees surrounded by agriculture lands and houses. At the border between Mexico and the U.S., the last of the Colorado's water is diverted, so that the last 160 kilometers of natural stream channel no longer carry regular flows and the river doesn't reach the sea. Because of this, the first binational environmental water allocation between Mexico and the United States was initiated (termed Minute 319) and two environmental flows were agreed upon.Those separate flow events provided an opportunity to investigate streambed infiltration under very different hydrographs, a “base flow” and a “pulse flow”. In this work, we evaluate the infiltration process using heat as a tracer and compare these point measurements to a saturated/unsaturated 2-D numerical model of a cross-section of the river. Heat tracer analysis showed that in both events, infiltration was high at the start and then decreased over time, from a maximum of 0.60 m d−1 during the pulse flow and 1.60 m d−1 during base flow. Monte Carlo analyses for the pair of temperature sensors at each event showed that uncertainty is greater at higher vertical fluxes; therefore greater during base flow (0.4 m d−1) and lower at pulse flow (0.2 m d−1). The total volume infiltrated during 6 and 8 days of the pulse and base flows (March 25-30 and September 7-14, respectively) was 151,714 m3 and 307,610 m3, respectively. In contrast, the 2-D numerical model showed 241,780 m3 and 125,295 m3 across the cross-section in a 500 m longitudinal distance of the river channel, during the pulse flow and base flow, respectively. In both models, the flux was highest during the initial days of each event. These results suggest that for future environmental flows focused on longitudinally reconnecting the river in conditions similar to those presented in this study, the minimization of channel transmission losses could be achieved through lower, consistent flows. However, this would not lead to reconnection of surface water and groundwater and therefore propagation of the wetting front laterally, which benefits restoration areas.