Relationships between riparian evapotranspiration and groundwater depth along a semiarid irrigated river valley

Evapotranspiration (ET) from riparian vegetation can be difficult to estimate due to relatively abundant water supply, spatial vegetation heterogeneity, and interactions with anthropogenic influences such as shallower groundwater tables, increased salinity, and nonpoint source pollution induced by irrigation. In semiarid south-eastern Colorado, reliable ET estimates are scarce for the riparian corridor that borders the Arkansas River. This work investigates relationships between the riparian ecosystem along the Arkansas River and an underlying alluvial aquifer using ET estimates from remotely sensed data and modelled water table depths. Results from a calibrated, finite-difference groundwater model are used to estimate weekly water table fluctuations in the riparian ecosystem from 1999 to 2009, and estimates of ET are calculated using the Operational Simplified Surface Energy Balance (SSEBop) model with over 200 Landsat scenes covering over 30 km² of riparian ecosystem along a 70-km stretch of the river. Comparison of calculated monthly SSEBop ET to estimated alfalfa reference ET from local micrometeorological station data indicated statistically significant high linear correspondence (R² = .87). Daily calculated SSEBop ET showed statistically significant moderate linear correspondence with data from a local weighing lysimeter (R² = .59). Simulated monthly SSEBop ET values were larger in drier years compared with wetter years, and ET variability was also larger in drier years. Peak ET most commonly occurred during the month of June for all 11 years of analysis. Relationships between ET and water table depth showed that peak monthly ET was highest when groundwater depths were less than about 3 m, and ET values were significantly lower for groundwater depths greater than 3 m. Negative sample Spearman correlation highlighted riparian corridor locations where ET increased as a result of decreased groundwater depths across years with different hydroclimatic conditions. This study shows how a combination of remotely sensed riparian ET estimates and a regional groundwater model can improve our understanding of linkages between riparian consumptive use and near-river groundwater conditions influenced by irrigation return flow and different climatic drivers.