Assessing Low Frequency Climate Signals in Global Circulation Models using an Integrated Hydrologic Model

Climate signals with periodicities of approximately one decade are pervasive in long-term streamflow records for streams in the western United States that receive significant baseflow. The driver of these signals is unknown but hypotheses have been presented, such as variations in solar input to the Earth, or harmonics of internal (i.e., processes in the ocean and troposphere) forcings like the Pacific Decadal Oscillation (PDO) and El Niño Southern Oscillation (ENSO). Climate signals of about 1 decade are important for several reasons, including their relation to climate extremes (i.e., droughts and floods), and because the drivers of these climate signals are clearly important for projecting future climate conditions. Furthermore, identifying the drivers of these climate signals is important for separating the relative impacts of human production of greenhouse gases on global warming verses external drivers of climate change, such as sunspot cycles. Studies using Global Circulation Models (GCMs) that do not incorporate solar forcings associated with sun spots have identified oscillations of about a decade long in certain model output. However, these oscillations can be difficult to identify in simulated precipitation data due to high frequency variations (less than 1 year) that obscure low frequency (decade) signals. We have found that simulations using an integrated hydrologic model (IHM) called GSFLOW reproduce decade-long oscillations in streamflow when driven by measured precipitation records, and that these oscillations are also present in simulated streamflow when driven by temperature and precipitation data projected by GCMs. Because the IHM acts as a low-pass filter that reveals low frequency signals (i.e. decadal oscillations), they can be used to assess GCMs in terms of their ability to reproduce important low-frequency climate oscillations. We will present results from GSFLOW applied to three basins in the eastern Sierra Nevada driven by 100 years of climate projection from six different GCMs. These GCMs will be evaluated according to their ability to simulate these important decade-long variations in climate by analyzing the power spectra of streamflow simulated by GSFLOW driven by the climate output from the GCMs.