CLIMATE CHANGE DATA PROJECTIONS FOR ONTARIO AND THE GREAT LAKES BASIN ECO Climate Data Roundtable, January 8, 2014 Richard Peltier, Physics, U Toronto
Regional Data Sets of Climate Change Projections 2 Provide climate change projections for Ontario and the Great Lakes Basin with an emphasis on the uncertainties concerning Future changes in extreme weather events (floods, droughts, heat waves, wind storms) Future changes in localized regions, eg. the Grand River Watershed of Southern Ontario We are especially interested in employing simulations of future climate to drive a detailed model of the surface and subsurface hydrology of individual regions in order to address the question of water resource availability impacts due to climate change Past funding: OME 2010-2012; IBM/SOSCIP 2012-2014
Dynamical Downscaling A2/RCP 8.5 business as usual trace gas scenario CCSM3: temperature change in 2050 relative to the 1950-1980 average Lake effect snow storm in, eg, Buffalo, NY Estimates of climate change on global scales Need future projections of climate change at the regional level if they are to be policy relevant
Methodology CCSM3/4 Output 3D nudging input files WPS Metgrid Geogrid Static fields Pressure-level Data on model grids Real Initialization Initial conditions, lateral and lower boundary conditions ARW MODEL WRF WRF Output System Flow Chart All analyses are being performed on 10 nodes of the IBM Power 6 parallel HPC platform
DYNAMICALLY DOWNSCALED CLIMATE CHANGE PROJECTIONS Global Climate Model (GCM) used for climate change projections on global scale IPCC AR5 summary for policymakers (2013) 100 kms Weblink to Dynamical Downscaling Movie 30 kms 10 kms Regional Climate Model (RCM) driven by GCM for responses to climate change at regional and local levels
Representation of Lake Influence WRF does not include an explicit accounting of lake influence but we may run a modern model of lake evolution either off-line or on line to properly represent this Freshwater lake model Flake [Mironov (2008)] -1D lake model intended as a lake parameterisation scheme for atmospheric models -two-layer parametric representation of temperature, heat and kinetic energy budgets (mixed-layer + thermocline) -Atmospheric forcing: T2, P, Q, SW, LW, U10, V10, snow fall rate Lake bathymetries are required: Global data set from [Kourzena (2009)]
Representation of lake influence CCSM3-Flake Simulations: Lake Irie (depth = 12m) Average ice duration (1979-2001) From Gula and Peltier, Journal of Climate, vol. 25, 7723-7742, 2012. Blue = FLake/CCSM3 simulations Black = Measurements from NDBC Buoy stations Gray = Ice observations from GLERL
Representation of lake influence CCSM-Flake Simulations: Lake Michigan (depth = 180m) Average ice duration (1979-2001) Blue = FLake/CCSM simulations Black = Measurements from NDBC Buoy stations Gray = Ice observations from GLERL
Historical period - Validation Lake-effect snow: Mean annual snowfall (1979-2001) NASA - Modis Snowbelt
Future scenario for Ontario Changes for 2050-2060 relative to 1979-2001over Ontario: WRF (Flake)
Changes in mean annual snowfall (%) for 2050-2060 relative to 1979-2001 Future scenario for Ontario
The importance of these snowbelt results as a demonstration of modern dynamical downscaling methods has been reinforced by the copy of the Gula and Peltier results in Alex Hall s (UCLA) recently (Dec 2014) published commentary in Science
UNCERTAINTIES IN CLIMATE CHANGE PROJECTIONS: Note: future climate change cannot to projected by physics free extrapolation http://vimeo.com/85531490 Weblink to multiple simulation Movie
To capture the uncertainty in future climate projections requires application of ensembles of simulations that differ from one-another in the representation of physical processes in the RCM and/or the initial conditions of the parent GCM integration and/or the GCM itself. This is a family of outer WRF domain temperature results from a mini-ensemble of WRF simulations (m,g,t,m,g) that differ only in the physical representation employed for certain critical processes in the RCM, with the parent GCM fixed to CESM1 with or without Arctic sea ice modification and for either mid-century or end-century END CENTURY!
CLIMATE AS WEATHER STATISTICS Extreme Value Analysis of daily precipitation From d Orgeville, Peltier Et al, JGR-Atmospheres,, 2014. Daily precipitation timeseries Tail distribution Return value for a given period of time 15 50 yr event 100mm/day 50 yr event 102mm/day 50 yr event 151mm/day 20 yr event 102mm/day but large uncertainties for one location and one simulation only
FUTURE PRECIPITATION CHANGES: Extremes OBSERVED (XX th century) SIMULATED (1979-1994) SIMULATED (2045-2060) 16 mm/day Precipitation amplitudes of 50 year events increase by 14 to 29% by mid-century Current 50 year events will occur more frequently every 25 to 15 years
FUTURE PRECIPITATION CHANGES: Annual averages OBSERVED (1979-1994) SIMULATED (1979-1994) SIMULATED (2045-2060) 17 mm/day Average precipitation increase 13 to 19% by mid-century Increase in extremes larger than increase of averages Precipitation increases are becoming more extreme
How extreme could the end-century climate of Ontario and the Great Lakes Basin become? These frames are for Future summer precipitation change over the WRF inner domain, ie Ontario and the Great Lakes Basin. Note the extreme summertime drying that is projected for end century conditions by model g. This is a consequence of the extreme exacerbation of drought conditions in the south western US that is a characteristic of this model du to its representation of land surface processes. Outlier Model g mm/day
Summary The dynamical downscaling methodology is able to provide robust estimates of future climate change at the regional scale which are useful for environmental policy development Robustness is achieved by employing ensembles of climate change projections that enable the construction of a probabilistic estimate of the expected change at a given future epoch in a given region The methodology requires the application of significant computational resources on the fastest computer systems available The Ontario government funded SciNet facility has been employed to demonstrate proof of concept in the application of this methodology to increasing understanding of the environmental future of the province. We need to develop an in-province technical capability to take full advantage of the existing facility and the upgrade to come in order to provide climate data services for the purpose of provincial policy development.
Historical period Validation Annual Winter Summer CCSM3 Annual mean precipitation (mm/day) for 1979-2001 WRF (10km) CRU Obs.
Historical period - Validation Tmax Annual Winter Summer Comparison of WRF downscaled results (10 km) over Ontario with Climate Research Unit (CRU)- University of East Anglia observational data for the 1979-2001 instrumental period Tmin WRF Vs CRU Obs. 1979-2001 CCSM Vs CRU Obs. 1979-2001 Tmax Tmin Annual Winter Summer
Future scenario for the lakes CCSM-Flake Simulations: Scenario SRES A2 Average ice duration (1979-2001) Average ice duration (2050-2060)
Average ice duration (1979-2001) Average ice duration (2050-2060)
Future scenario for Ontario Changes in number of days with snow for 2050-2060 relative to 1979-2001:
Connecting the future climate projections To changes in water resource availability HydroGeoSphere is a threedimensional control-volume finite element simulator which is designed to simulate the entire terrestrial portion of the hydrologic cycle. This research is being conducted in collaboration with the hydrology group of Ed Sudicky of U waterloo and Acquanty
7000 km 2 Population of ~900,000 Intensive Agriculture 93% rural/agricultural land use 290,000 head of cattle 500,000 thousand swine 8.8 million poultry 900 mm of precipitation/year Heavy Dependence on Groundwater for Municipal Water Supply Heavily Instrumented Long Term Records (Image courtesy of Grand River Conservation Authority) 26
Surface layer Topography Land use, surface water channels Soil types, hydraulic characteristics Surface domain Subsurface 15 hydrostratigraphic units 3D contact surfaces Interactions Subsurface domain 27
Observed vs. Simulated Surface Drainage Network Depth [m] Observed Drainage Network Simulated Surface Water Depth 28
Observed vs. Simulated: Stream Flow and GW Head Observed vs. Simulated Stream Flow Observed vs. Simulated GW Head 29
Changes in Annual Total Precipitation: 12 % Increase by Mid-Century; 14 % Increase by End-Century Relative Change Relative Change Relative Change Historic to Mid-Century Mid- to End-Century Historic to End-Century 30
Changes in Annual Total Potential Evapotranspiration: 8 % Increase by Mid-Century; 22 % Increase by End-Century Relative Change Relative Change Relative Change Historic to Mid-Century Mid- to End-Century Historic to End-Century 31
Changes in Steady-State Depth to Water Table Depth to WT Change [m] Historic to Mid-Century Mid- to End-Century Historic to End-Century 32