A critical hydrologic process in arid and semiarid regions is the interaction between ephemeral channels and groundwater aquifers. Generally, it has been found that ephemeral channels contribute to groundwater recharge when streamflow infiltrates into the sandy bottoms of the channels. This process has traditionally been studied in channels that drain large watershed areas (10s to 100s km2). Since the water table in arid and semiarid regions may be 10s to 100s of m below the land surface, measured streamflow losses or percolation into the deep vadose zone is assumed to eventually lead to groundwater recharge. In this study, we use a water balance approach to estimate deep percolation in a first-order, instrumented watershed (4.6 ha) in the Jornada Experimental Range (JER). Observations at the JER have shown that over the last 150 years vegetation has changed dramatically from grassland with high vegetation cover to a shrubland with large amounts of bare soil, a transition which should affect deep percolation.
Our results indicate that runoff generated in first-order watersheds can contribute significantly to deep percolation. During our ~6 year study period, we observed 428 mm of percolation and a ratio of percolation (P) to rainfall (R) of P/R = 0.27. Utilizing our dense instrument network, we determine that this percolation occurs inside the channel areas themselves when these receive runoff from upland hillslopes. We observe much less runoff (Q) at the watershed outlet than the estimated percolation, there were only 39 mm of runoff during the study period, which is a ratio of runoff to rainfall of Q/R = 0.02. Using the long term data from the JER, we then estimate that 48 mm of percolation per year occurred on average at our study site and we discuss the historical trends in estimated percolation. These trends are related to changing rainfall patterns, though it is likely that the encroachment of woody shrubs, which increases bare soil and facilitates movement of water into channels, would also play a role in the actual percolation. Additionally, we identified >1,700 similar small (4-5 ha), first-order watersheds located in the JER that potentially also contribute upland percolation for groundwater recharge. Our results highlight the importance of first-order shrubland watersheds in producing groundwater recharge and should help improve water balance quantifications and overall recharge estimates in arid shrublands.