Determining the Local Implications of Global Warming

Determining the Local Determining the Local Implications of Global Warming Implications of Global Warming

Clifford Mass and Eric Salathe,

Patrick Zahn, Richard Steed

University of Washington

Project Support

• King County

• Seattle City Light

• EPA STAR Program

• Climate Impacts Group

Questions

What are the implications of global warming for the Northwest?

How will our mountains and land-water contrasts alter the story?

Do global models tell us the full story?

Regional Climate Prediction

• To understand the impact of global warming, one starts with general circulation models (GCMs) that provide a view of the global evolution of the atmosphere.

• GCMs are essentially the same as global weather prediction models but are run with much coarser resolution and allow the composition of the atmosphere to vary in time (e.g., more CO2)

• Even leading GCMs only describe features of roughly 500 km or larger in scale.

•Northwest weather is dominated by terrain and land-water contrasts of much smaller scale.•In order to understand the implications of global changes on our weather, downscaling of the GCM predictions considering our local terrain and land use is required.

Model Topography and ResolutionMM5 Topo (15 km) ECHAM5 Topo (150km)

Annual Precipitation

Downscaling

Downscaling

• The traditional approach to use GCM output is through statistical downscaling, which finds the statistical relationship between large-scale atmospheric structures and local weather.

• Statistical downscaling either assumes current relationships will hold or makes simplifying assumptions on how local weather works.

Downscaling

Such statistical approaches may be a reasonable start, but may give deceptive or wrong answers… since the relationships between the large scale atmospheric flow and local weather might change in the future.

Downscaling

• There is only one way to do this right… running full weather forecasting models at high resolution over extended periods, with the large scale conditions being provided by the GCMs….this is called dynamical downscaling.

• Such weather prediction models have very complete physics and high resolution, so they are capable of handling any “surprises”

Example of Potential Surprises

• Might western Washington be colder during the summer under global warming?– Reason: interior heats up, pressure falls,

marine air pushes in from the ocean

• Might the summers be wetter?– Why? More thunderstorms due to greater

surface heating.

Downscaling• Computer power and modeling

approaches are now powerful enough to make dynamical downscaling realistic.

• Takes advantage of the decade-long work at the UW to optimize weather prediction for our region.

UW Regional Climate Simulations

• Makes use of the same weather prediction model that we have optimized for local weather prediction: the MM5.

• 10-year MM5 model runs nested in the German GCM (ECHAM).

• MM5 nests at 135 km, 45 km, and 15 km model grid spacing.

MM5 Model Nesting• 135, 45, 15 km MM5 domains• Need 15 km grid spacing to model local weather features.

Regional Modeling

• Ran this configuration over several ten-year periods:

• 1990-2000-to see how well the system is working

• 2020-2030, 2045-2055, 2090-2100

Details on Current Study: GCM• European ECHAM model with resolution roughly

equivalent to having grid points spaced ~ 150 km apart. Can resolve features of roughly 600 km size or more.

• IPCC climate change scenario A2 -- aggressive CO2

increase (doubling by 2050)

IPCC Report, 2001IPCC Report, 2001

Global Forcing: Surface Temperature

First things first

• But to make this project a reality we needed to conquer some significant technical hurtles.

• Example: diagnosing and predicting future deep soil temperatures

• Example: requirements for acquiring GCM output every 6 h and storing massive amounts of output.

• Evaluating the 1990-2000 simulations

Evaluating of Model Fidelity

• We have carefully evaluated how well the GCM and the MM5 duplicated the 1990-2000 period.

• We previously had run the system using another GCM…the Parallel Climate Model…with unsatisfactory results….crazy cold waves during the winter.

• ECHAM Model appears far better…but not perfect.

Too Cold

• Cold episodes occurred 1-2 times per winter with temperature getting unrealistically cold (below 10F) in Puget Sound:

• Also a general cold bias to minima

• Better than previous attempts.

Why Cold Outbreaks?

• Widespread surges of arctic air originate in ECHAM5, likely owing to poorly-resolved terrain (Cascades and Rockies).• Extreme cold air inherited by MM5.• Results from previous experiments with lower-

resolution (T42) GCM indicate that higher resolution reduces frequency and severity of unrealistic cold events.

• Also problem in model physics--probably more important

The Fix

• Our research during the past few months suggests the problem was a bug in the land surface model.

• Fixed in the current version and will be used in next production runs.

Evaluation of Future Runs

Because there are some biases in the GCM runs, results for future decades (2020s, 2040s, and 2090s) will be evaluated against the ECHAM5-MM5 1990-2000 baseline

Now, The Future

Why Such Strong Warming on Mountain Slopes..Particularly in

Spring?

• Probable Answer: Snow melt resulting in more solar heating.

Change inWaterOfSnowpack(%)

Solar Radiation

Snow and Ice Reflect Much of The Incoming Solar Radiation

Now

Solar Radiation

Global Warming Causes Snow level to Rise Resulting In Absorption of Solar Energy on Melted Slopes

=WARMINGFuture

Why Cooling West of Cascades in Spring?

• Low clouds due to more onshore flow from off the cool, cloud Pacific.

• The Montereyization of the western lowlands!

Precipitation

• Bottom Line: No Large Regional Trends

Summary

• The viability of the approach…using high resolution numerical prediction models forced by large-scale general circulation climate models (GCMs)… has been demonstrated.

• Careful evaluation of the GCM output is required…there are deficiencies.

• Although there is general warming over the region for all seasons, the terrain and land water contrasts of the region enhance or weaken the warming in certain areas.

Summary• Warming is enhanced on the upper windward slopes

due to snow melt.• Springtime warming is lessened west of the Cascade

crest due to more low clouds.• Many more hot days during the summer.• Precipitation changes are more modest then

temperature changes.• There will be a substantial loss of snowpack,

reaching catastrophic decreases by 2090.

Future Work

• We are just in the beginning of this work.• Need to find and remove causes of biases and cold

outbreaks• Need to test other global warming scenarios• Will try to find higher resolution GCMs• Try more sophisticated MM5 physics• More analysis.

The END

• :

Climate Change in the Pacific Northwest:

Do Global Models Tell the Whole Story?

Climate Change in the Pacific Northwest:

Do Global Models Tell the Whole Story?

Eric SalathéCSES Climate Impacts Group

University of Washington

With: Cliff Mass, Rick Steed, Patrick ZahnWSU, USDA Forest Service, NCAR

Eric SalathéCSES Climate Impacts Group

University of Washington

With: Cliff Mass, Rick Steed, Patrick ZahnWSU, USDA Forest Service, NCAR

IPCC Scenarios for Pacific Northwest Climate Change

IPCC Scenarios for Pacific Northwest Climate Change

Empirical Downscaling• Assumes climate model captures temperature and precipitation trends

Regional Climate Model•Represents regional weather processes• May produce local trends not depicted by global models

Downscaling Methods Used in CIG Impacts studies

Mesoscale Climate Model -- MM5

Based on MM5 Weather Model

ECHAM5 Climate Model used to force Mesoscale Simulation

Nested grids 135-45-15 km

Nudging on outermost grid by forcing global model

Advanced land-surface model (NOAH) with interactive deep soil temperature

More RainMore Rain

Shift in Pacific Storm Track

Salathé, Geophys Res Lett, 2006

NCEP-NCAR Reanalysis

20th Century Model Composite

21st Century Model Composite

1950-2000 to 2050-2100 Nov-Dec-Jan

Composite of 10 Global Models

“Observed” Climate

MM5 Result for Sep-Oct-Nov

Change in Sep-Oct-Nov Precip (mm/day) 1990s to 2050s

Contours:Change in500-mb heights

IncreasedWesterlyFlow

MM5 vs Statistical DownscalingStatistical Downscaling

Precip only Precip & Winds MM5

Change in November Precip (mm/day) 1990s to 2050s

More WarmingMore Warming

Winter Warming in MM51990s to 2050sTemperature Change

Difference betweenMM5 and ECHAM5

Change in Winter Temperature (degrees C)Change in Winter Temperature (degrees C)

Less warming In MM5

More warming In MM5

Loss of Snow cover and WarmingTemperature Change

Change in Winter Temperature (degrees C)

Snow Cover Change

Change in fraction of days with snow cover

January to April in MM5

-50

Comparison of MM5 and HCN Observations

Consistent trend over 21st Century

2020s 2050s 2090s

Change in Winter Temperature (degrees C)

Winter Trends at Various StationsMM5 - ECHAM5

Te

mp

era

ture

Ch

an

ge

(°C

)

Winter Trends at Various StationsMM5 - ECHAM5

Te

mp

era

ture

Ch

an

ge

(°C

)

10 IPCC Models

1950 2000 2050 2100

Regional Model Compared to Global Model

2020s 2050s 2090s

Change in Winter Temperature (degrees C)