How what we know about climate projections translates into hydrology - PowerPoint PPT Presentation

How what we know about climate projections translates into hydrology projections and water resource decisions for Tampa Bay Water 2018. 4. 5 Seungwoo Jason Chang, Water Institute, University of Florida Wendy Graham, Water Institute, University of Florida

Introduction The FloridaWCA, UF Water Institute & Tampa Bay Water Goal: To increase the regional relevance and usability of climate and sea level rise models for the specific needs of water suppliers and resources manages in Florida. Tampa Bay Water Project Research objectives: Evaluate impact of future climate scenarios on future water supply availability in the Tampa Bay region.

Framework Long Term Water Resources Projection Analysis Framework: Precipitation Climate Temperature Solar radiation Evapotranspiration Hydrologic Long-term simulation Impact water (Regional assessment resources hydrologic Public pumping planning models) Human Ag. pumping impacts Irrigation Land use change

Framework Long Term Water Resources Projection Analysis Framework: Don’t forget bias correction! Dynamical High Downscaling GCMs resolution (low climate for resolution) regional Statistical study Hydrologic Downscaling Long-term simulation Impact water (Regional assessment resources hydrologic Public pumping planning models) Human Ag. pumping impacts Irrigation Land use change

Projects Ok! Dynamic downscaling of coarse climate data. What we did? - Used MM5 to dynamically downscale precipitation from NCEP-NCAR reanalysis data. Why we did it? - To test the accuracy of dynamically downscaled climate model to reproduce climate variables at scales needed for regional retrospective hydrologic studies. What we found? - Significant errors (daily P) are found even after bias-correction, maybe ok for multi-decadal water resource planning - We should leave climate modeling to the climate modelers! Hwang et al. (2011), Journal of Hydrometeorology

Projects Ok! Dynamically downscaled climate data for regional hydrologic study. What we did? - Used FSU’s dynamically downscaled retrospective climate data to simulate streamflow. Why we did it? - To test the ability of dynamically downscaled retrospective climate data to reproduce retrospective hydrology What we found? - Bias correction is required to obtain reliable hydrologic predictions. Hwang et al. (2013), Reg. Env. Change

Projects Acceptable… Comparison of dynamically downscaled reanalysis data What we did? - Compare four dynamically downscaled climate data to simulate streamflow and GW. Why we did it? - To investigate how differences in dynamically downscaled climate data propagate into hydrologic predictions What we found? - All products had errors that were propagated and enhanced by hydrologic models, results OK All four have timing issues and magnitude issues for multi-decadal planning Hwang et al. (2014), Journal of Hydrology

Projects Ok! Development of statistical downscaling method (BCSA) What we did? - Developed a new statistical downscaling method. Why we did it? - Existing statistical downscaling methods did not reproduce rainfall characteristics in FL very well. Dynamic downscaling is computationally intensive What we found? - Choice of statistical downscaling method matters in FL. Small-scale spatial variability is important. Hwang and Graham (2013), Hydrology and Earth System Sciences

Projects Acceptable… Comparison of downscaling methods What we did? - Evaluated hydrologic implications of statistical downscaling methods. Why we did it? - To understand possible hydrologic implications of different statistical downscaling methods. What we found? - Choice of how you translate global model output to finer spatial scales matters for water resources planning. SDBC and BCSA OK SDBC and BCSA ok! Hwang and Graham (2013), Journal of the American Water Resources Association

Projects Also important! Sensitivity of future water deficit projections using GCMs Annual mean change in P-ET 0 (mm day -1 ) What we did? - Evaluated the sensitivity of future water deficit projections to GCM, ET 0 method and RCP selection Why we did it? - To understand sources of uncertainty when using climate projections for future water resources planning. What we found? - For Southeast US, GCM uncertainties and ET 0 methods uncertainties are both important. Chang et al. (2016), Hydrology and Earth System Sciences

Simple bias correction is good enough! Projects Univariate bias correction vs Joint bias correction What we did? - Compared the performance of two bias correction methods to reproduce correlation among hydrologically important climate variables (P and ET 0 ) and predict regional hydrologic response. Why we did it? - To determine most appropriate bias correction method for Tampa Bay Water region. What we found? - For TBW, simple sequential univariate bias correction was satisfactory for water resources planning . Chang et al. (In progress)

Projects Univariate bias correction vs Joint bias correction: What about rest of USA? Illinois Water Supply North Carolina Orange Water and Sewer Authority

Projects Univariate bias correction vs Joint bias correction P vs ET 0 show better performance than P vs T P vs ET 0 P vs Tmax Joint bias correction is better. Possible to use simple univariate bias correction. Chang et al. (In progress)

Projects Ok Climate change vs anthropogenic change What we did? By human change - Evaluated future hydrologic projections resulting from alternative climate change and human water use scenarios. No significantly different. Why we did it? - To understand the relative importance of changes in climate By GCM versus human water use for projecting future water supply What we found? Significantly different. - Differences among climate projections most significant for streamflow projections, but differences among human water use scenarios are also significant for GW projections. Chang et al. (Under review), Hydrology and Earth System Sciences

Projects Dynamical vs Statistical Downscaling methods. 8 8 RAW-CCSM4 RAW-CCSM4 LIVNEH LIVNEH 7 7 BCSA BCSA ROMSC ROMSC What we did? ROMSU ROMSU 6 6 BC-ROMSC BC-ROMSC BC-ROMSU BC-ROMSU - Compare FSU’s new dynamically downscaled Precipitation (mm/day) Precipitation (mm/day) 5 5 climate data to statistical downscaled climate to 4 4 see if it improves regional hydrologic predictions. 3 3 2 2 Why we did it? 1 1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 - To take advantage of recent advances in Month Month 1.5 1.5 dynamic downscaling methods (coupled 1 ocean-atmospheric regional climate models) 1 0.5 0.5 Change in precipitation (mm/day) Change in precipitation (mm/day) What we found? 0 0 - Bias correction is still required and … -0.5 -0.5 CCSM4 RAW-CCSM4 BNU-ESM BCSA GFDL-CM3 ROMSC -1 GFDL-ESM2G -1 ROMSU MIROC-ESM BC-ROMSC MPI-ESM-LR BC-ROMSU MRI-CGCM3 -1.5 -1.5 NorESM1-M 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 bcc-csm Month Month

Projects What have we learned (big picture)? • GCMs predict a consistent increase in temperature for Florida (1-3 o C for 2040-2070) • Future GCM precipitation projections vary widely for Florida and these differences propagates into significantly different hydrologic projections • Downscaling and bias-correction approach matters. Bias correction is always important • Need to use multiple GCMs in any future water resource planning efforts and look for robustness of plans across wide range of projections.

Future plan Future plan: Water resources planning for Tampa Bay Water Precipitation Climate Temperature Solar radiation Hydrologic Long-term simulation Evapotranspiration water Impact (Regional resources assessment hydrologic planning models) Public pumping Human Ag. pumping impacts Irrigation HOW to link all information we have? 1. Hydrologic projections. Land use change 2. Potential new supply projects. 3. Decision triggers? Optimization

Thank you Seungwoo Jason Chang: swjason@ufl.edu