Impact of Climate Change on Surface and Subsurface Water Interaction and Riparian Vegetation: Linkage between Hydrology and Invasive Tamarisk in a Semi-Arid Basin
AuthorBhattarai, Mahesh Prasad
StatisticsView Usage Statistics
Long-term assessment of climate change impacts on water resources is essential for developing management strategies. They are even more important in semi-arid basins of the southwestern United States where effects of changes in precipitation and temperatures are likely to be intensified by the semi arid nature of the basin. Proliferation of tamarisk (salt cedar) has negatively impacted the ecology and hydrology of riparian ecosystems in the southwestern United States. Riparian systems are undergoing perturbations in hydrology from anthropogenic factors such as aquifer pumping, climate change, and impoundment. In this study, the Lower Virgin River Basin (LVRB) in southeastern Nevada, southwestern Utah and northwestern Arizona was used to evaluate hydrologic response of a semi-arid basin under current and future climatic conditions. Hydrologic simulation was performed using a fully integrated physically based surface-subsurface model, Hydrogeosphere (HGS). Hydrologic parameters derived from the simulations and bioclimatic variables were then used to develop a species distribution model (SDM), Maxent, to evaluate future competitiveness of tamarisk, and its success or failure under Special Report on Emission Scenarios (SRES) A2. The LVRB subsurface system is comprised of many aquifers, but to meet research objectives, the focus of the study was limited to the upper aquifers (channel-fill deposits, Tmc1 and Tmc2). These upper aquifers are in direct contact with surface water, support riparian vegetation and are the source for most groundwater withdrawal in the basin. The simulation period for current conditions spanned from 2007 to 2010, in which 2007 to 2008 was used for calibration and 2009 to 2010 for validation. The model simulation results showed that stream hydrographs were similar to observed peak flows and dry weather flows. The Pearson correlation coefficients for Littlefield and near Overton were 0.93 and 0.81, respectively, and Nash-Sutcliffe efficiencies were 0.95 and 0.41, respectively, for the simulation period 2007 to 2010. However, both stations were negatively biased due to low subsurface discharge to the Virgin River. The model was useful in identifying losing and gaining sections of the river. Statistical errors calculated for a continuously monitored well showed that mean residual error (MRE), mean absolute residual error (MARE) and root mean square error (RMSE) were 0.71 m, 1.10 m, and 1.37 m, respectively. Comparison of simulated and actual ET for 2010 at the Virgin River meteorological station yielded correlation coefficient (R2) and RMSE of 0.98 and 0.50 mm/d, respectively for the period. In general, simulations were good and use of a coupled, physically based model helped to identify losing and gaining sections of recharge and discharge areas, runoff generation, and spatial and temporal distributions of actual ET. Long-term assessment of climate change impacts was evaluated under SRES A2 climatic projections for the last four years of 21st century, 2096-2099. Climate projections data from three Global Climate Models (GCM), ECHAM5 (Max-Planck Institute for Meteorology), HADCM3 (UK Hadley Center for Climate Prediction and Research), and CCSM3 (National Center for Atmospheric Research) were used. The Virgin River is a losing river in the lower basins. River loss increased by 12.5% for the last four years of 21st century compared to 2007 2010. The percent recharge, in terms of total input flux, decreased from 4.4% to 3.4% from 2007 2008 to 2096 2099. Simulated actual ET for 2099 at the Meadowland Farm Station showed an increase of 17.8% compared to that of 2010. Water level in six wells dropped from 0.94 m to 6.5 m for the same period. Overall, the LVRB will likely experience increase in ET, decrease in recharge and lowering of groundwater table under SRES A2 conditions. The species distribution model, Maxent, was used to predict future tamarisk coverage with hydrological data such as depth to ground water and soil moisture content for current climate conditions and future climate projections generated from HGS simulations along with bioclimatic variables. Three separate Maxent models (M1, M2 and M3) were developed, first (M1) with bioclimatic, elevation and soil variables, second (M2) with bioclimatic, hydrologic, elevation and soil variables and third (M3) with all the variables in second model but without temperature related bioclimatic variables. Modeling results with the combinations of different variables revealed that hydrologic variables, especially the depth to groundwater are the most important predictors of tamarisk distribution, especially for the months of February to April. M1 model only with bioclimatic variables estimated higher environmentally suitable area than M2 model with groundwater depth variables and soil saturation variables. M2 model showed likely increase tamarisk proliferation area close to the river channels, especially along the VR and likely retreat from the headwater areas of BDW. M3 models without temperature variables projected the least suitable areas. Trend of the model results indicated that inclusion of temperature change in the model contributes toward increasing environmentally suitable areas compared to the models without it.