Characterizing changes in hydrologic fluxes and uncertainties for current and future climate conditions

Abstract

The global mean surface temperature has increased by roughly 1 ⁰C during last century due to the greenhouse gas emission. According to the Fifth Assessment report of Intergovernmental Panel on Climate change (IPCC AR5), the surface temperature is projected to likely increase by 0.3-4.8 ⁰C under different emission scenarios, compared to 1986-2005. Although the surface temperature is likely to increase globally, changes in precipitation are projected to be spatially non-uniform leading to potentially significant changes in the hydrologic cycle for select regions. To understand how the hydrologic cycle may be impacted by altered climate conditions, the objectives of this work are to: (i) investigate the changes in hydrologic fluxes under current and future climate conditions, (ii) identify the main driving mechanisms of these changes, and (iii) quantify the associated uncertainties for the estimated impacts. This project investigated the spatial and temporal variations of streamflow in multiple regions (eastern and southwestern United States, western Amazon) under altered climate conditions. A hydrologic modeling framework which integrates multiple runoff generation approaches with surface, subsurface and channel routing processes was developed and an auto-calibration routine was included. Forced with hindcasted and projected temperature and precipitation from a select suite of the General Circulation Models (GCMs) in the Coupled Model Intercomparison Project Phase 5 (CMIP5), this framework enables the estimation of streamflow and corresponding uncertainties resulting from different sources, which is essential to extract actionable information for stakeholders regarding adaptation to climate change. The primary findings from this work reveal a non-linear hydrologic response to altered precipitation, which leads to larger variability and pronounced changes in streamflow. The non-linear response implies that changes in precipitation (seasonality, magnitude and timing) result in even larger changes in streamflow (peaks and total runoff), especially for the extremes. The estimated changes in streamflow show large uncertainties mainly due to model forcings (i.e., the GCMs and RCPs) accounting for nearly 50% of the total uncertainty. Different runoff generation process representations account for an additional 10-15% of the total uncertainty while the choice of hydrologic model parameter sets results in minor contributions to the total uncertainty.