The Best Estimate Yet of the Impact of Global Warming on the Pacific Northwest

During the past month, a group of atmospheric researchers in my group at the University of Washington have completed a series of high-resolution climate simulations that probably represent the best estimate to date of what global warming, forced by increasing greenhouse gases, will mean for the Northwest.

CO2 is increasing more rapidly then ever

This blog will explain what we have done and how it provides a superior tool for determining the regional implications of anthropogenic (human-forced) climate change.  Securing a realistic estimate of the local impact of global warming is critical if we are to take steps to ensure resilience to the expected changes.  And yes, it might motivate folks to reduce their carbon footprint (like reducing jet travel, using mass transit more, etc.)

Before I begin, let me note that much of this work was sponsored by an Amazon Catalyst grant, which supported two UW staff members (Richard Steed and Jeff Baars) who ran and analyzed the simulations.   Amazon also provided substantial cloud computing resources.

Most estimates of climate change begin and end with global climate models, which include both atmospheric and ocean modeling components.  These global models are driven by estimates of how greenhouse gases, like CO2, will increase in the future.

But there is a problem.  To run globally for a century or more, the climate models must sacrifice spatial resolution.  In fact, most global models are run with grid spacing of around 150 km, which makes them unable to simulate the impacts of the crucial terrain and land-water features of the Pacific Northwest.  

We are talking about no Cascades, no Olympics, no Puget Sound or Strait of Juan de Fuca.  Even the Rockies are too low and smeared out.   Not good.  One thing my group at the UW has shown is that you have to run weather prediction models with roughly 12-km or better grid spacing to have any chance of getting our local weather and climate correct. The picture below illustrates the difference in terrain between a climate model (e.g., NCAR's CCSM4) and a higher-resolution weather prediction model.

A number of research projects have only made use of one or two climate models.  But each model has its uncertainties and biases; it would be far better to look at a collection (or ensemble) of model results--just as it is better to look at an ensemble of weather prediction forecasts.

The UW Effort:  An Ensemble of High-Resolution Climate Simulations

To address these issues, our group took on a complex task:  to run a high resolution weather forecast model for 130 years (1970-2100).   But instead of running the high-resolution model globally, which would have required impossible computer resources, we ran it over a domain that includes only the Pacific Northwest (see domains above).  And we did this 12 times, each driven by a different climate global model, allowing us to get a handle on the uncertainty in the climate forecasts.  There are fancy names for what we did:  dynamical downscaling or regional climate modeling.

We started with the several dozen global models that were part of the international CMIP-5 effort and selected the twelve that verified the best over the eastern Pacific during a contemporary period (1970-2000) and had output every six hours (which we needed to drive our regional climate model).

But we had a major decision to make.  How aggressive an increase in greenhouse gases should we use?   Should we assume that folks would "get religion" about climate change and radically reduce their emissions or should we use a "business as usual" continuation of recent trends?

We decided to go with the latter, in order to delineate the worst case.  So we used the RCP 8.5 scenario, which assumes continued increases of emissions (see below).   For the next few decades, the emission scenarios are all similar, since

 mankind can't change our energy technologies quickly, but by the end of the century the differences are large. 

Let me say, that I believe that the RCP 8.5 scenario will turn out to be too pessimistic (aggressive) by the middle to the end of the century.  I suspect that advancing technologies (like better photovoltaics, the advent of fusion power, advances in removing CO2) will have a major impact 30-60 years from now in reducing CO2 emissions and concentrations.   There will be advances that we can not even imagine now.

So consider what I am about to show as the worst case---and that the actual changes will not be so extreme.

Some Results

I will start by showing you the difference in conditions between a recent period (1970-1999) and two future periods (2030-2059, 2070-2099).    I show such thirty year periods to average out short-term variations that have little to do with climate trends. And I will present averages of the 12 climate runs.

Looking at the annual average of maximum temperatures, there will be roughly a 2C (about 3.6F) warming in maxima by 2030-2059--think 2045 (click on figures to enlarge).

But by the end of the century (2070-2099, think 2085), the average over the region will be twice that (around 4C increase, about 7F), with considerably more warming in the interior.

What about precipitation?  Looking to the first period (2030-2059), there is a generally small increase in precipitation, but there is something unexpected and subtle---decreases in precipitation in the downstream of some barriers.  An interesting finding to investigate further.

By the end of the century, the precipitation increases are more dramatic.   The Northwest will not dry out under global warming--we will generally get wetter.

Now, let me show you a plot of the annual mean temperatures from all ensemble members at a particular location (in this case Seattle).  Observed temperatures are shown with black dots.

The model runs are reasonably close to the observations before 2018, perhaps a degree too warm.  You will notice a steady rise over the century--nothing abrupt.  It appears that global warming might have contributed about a degree (C) of warming since 1970, with an additional 4C by the end of the century.

Summer precipitation at Seattle?  Lots of variability, but it looks like there will  be a small drying (perhaps 1 inch over the summer) by the end of the century.

Winter precipitation in Seattle?  A small increase, again with lots of year to year variability.

Want to see something scary?  Next, let's look at changes to our snowpack.

Here is the are the April 1 snowpack for 1970-1999, 2030-2049, and 2070-2099.  Modest declines over the lower-elevation terrain by mid-century.  Less change on the high terrain in British Columbia.  Much larger declines by the end of the century.  I don't think there will be skiing in Snoqualmie Pass in 2085 if the warming is not reduced.

Summer winds speed changes at the surface?  Generally weaker, except in the Strait of Juan de Fuca and Strait of Georgia.  Not good for wind generation.

The results I have shown you above are just the tip of the iceberg on what we can explore with these model runs.  Will cloudiness change?  How about the strong winds that drive wildfires?  There are many interesting questions that can be addressed with such climate simulations.

Major Lessons

The runs described above represent the best guidance now available for how global warming will influence our region if nothing is done to address greenhouse gas emissions.   By the end of the century, there will be substantial warming, with the average summer day around Puget Sound climbing into the mid-80s, rather that the upper 70s of today.  Our typical winter day west of the Cascades will have a high around 50F.   Cascades snowpack will decline substantially on April 1st (about 50% below today's value at 5000 ft). Winters will be wetter, but summers slightly drier (they are already typically dry today, but will be even more so under global warming).  Temperatures below freezing will become rare by the end of the century here in Seattle.  East of the Cascades, the influence of warming will be greater (see example for Yakima minimum temperatures below)

We are already seeing some small  temperatures impacts of global warming (1-2F), which implies that the major heat waves  TODAY are mainly natural variability (if temperatures is 20F above normal, as we observed earlier this month, roughly 18F of that warmth is natural).

Next Steps

There is much that should be done next.  First, we need to statistically improve our current runs, using bias correction based on the contemporary period. Next, we should run these simulations with an improved version of our modeling systems, using less aggressive and probably more modest global warming scenarios.  We should run our regional model driven by the next generation of global climate models (CMIP-6) and add physics variability and different start dates to get a wider range of solutions.

Unfortunately, our Amazon funds have been expended....so if you represent a foundation or a potential donor, and are interested in helping, please let me know or check here. 

Change in summer precipitation in inches

I believe simulations like those shown above will provide major assistance to a society that will have to adapt to a certain amount of climate change.  I also think that someone could make a good business out of providing regional climate prediction services.   But this not the kind of thing that the National Science Foundation would support (too applied) and the State of Washington doesn't support much outside climate research.