ModelIng & Analysis

Our understanding of current and projected climate impacts results from our NOAA RISA team’s efforts to advance the state of the science in modeling and analysis. This includes the model evaluation work we did for our Integrated Scenarios project, which has proven to be a key factor in creating better local climate projections. It has also contributed to the ongoing conversation in the Pacific Northwest’s climate adaption and scientific community about how best to apply data from global climate models to the modeling of local climate impacts. CIRC also developed an advanced a method for downscaling climate model data to local levels, used distributed computing to hone the use of a regional climate model, created a simplified approach for understanding future changes in watersheds, and created a modeling framework to determine the cumulative impact of multiple coastal hazards.

Global Climate Model Evaluation

As part of CIRC’s Integrated Scenarios project, our NOAA RISA team members, led by CIRC researcher David Rupp, performed an evaluation of data output from global climate models associated with the Fifth Phase of the Coupled Model Intercomparison Project (CMIP5). The goal of the evaluation was to determine which models produced outputs that were the best statistical fit for the Pacific Northwest. This was done by testing how well the models were able to simulate various measures of the region’s historical climate. The idea was that models that faithfully reproduced the Pacific Northwest’s past climate would also produce the highest fidelity simulations of the region’s future climate.

Visualizing GCM Evaluation  


  • Climate model evaluation work done for Integrated Scenarios. Models are listed at the bottom. On the left are various measures for examining temperature and precipitation. The graph depicts what’s known as relative error, in this case how well the models match historical measures for the Pacific Northwest. Here warm colors depict higher degrees of error and cooler colors less error. The models are organized from left (least error) to right (most error). (Image: David Rupp, “Evaluation of CMIP5 20th century climate simulations for the Pacific Northwest USA”, Journal of Geophysical Research: Atmospheres, VOL. 118)


  • Learn More about Climate Models via the Northwest Climate Toolbox 
  • Presentation by Dr. David Rupp recorded on April 17, 2014 at workshop in Portland, Oregon, for the Integrated Scenarios of the Future Northwest Environment project, an effort to understand and predict the effects of climate change on the Pacific Northwest's fish, wildlife, hydrology, and ecosystem services.

2 Integrated Scenarios: Climate Change Projections for the Northwest



  • Data from global climate models (GCMs) that were the best statistical fit for the Pacific Northwest—that were best able to simulate the region’s historical climate—also projected the most warming for the region under climate change (Rupp et al. 2013).


  • This model evaluation effort fed CIRC adaptation efforts, including the Piloting Utility Modeling Applications (PUMA) and Integrated Scenarios projects, and has garnered over 100 citations in peer-reviewed studies.



  • Rupp, David E., John T. Abatzoglou, Katherine C. Hegewisch, and Philip W. Mote.
    “Evaluation of CMIP5 20th CenturyClimate Simulations for the Pacific Northwest USA.” 
    Journal of Geophysical Research: Atmospheres 118, no. 19 (2013).

  Thinking Deeper:

About Global Climate Models

As their name suggests Global Climate Models (GCMs) model the whole Earth. They do this by dividing the globe into large three-dimensional, box-like cells that include the atmosphere to the space just above the ground. Inside these cells atmospheric dynamics are simulated within the computer. As with a video game or other computer simulation, the cells have a resolution that’s dependent on how much information can be packed into a given cell. For GCMs, the more cells you have to divide the world into, the more detailed and higher resolution that simulated world will be. (Incidentally, this isn’t just a metaphor; “resolution” is the term of art in the climate-modeling world.)

Yet, while engineers have consistently built higher and higher resolution into GCMs—thank you, Moore’s law—this resolution still isn’t fine enough to assess regional climate impacts in any great detail. Consider an important Pacific Northwest feature: the Cascade Mountains. Under the coarse resolutions of many GCMs, the Cascades are replaced by a kind of featureless plain that gently tilts upward toward Wyoming. This lack of detail produces a problem for climate researchers. The Cascades create a major rain shadow that makes the eastern half of the Pacific Northwest significantly drier than its western counterpart. That rain shadow needs to be accounted for. Detail needs to be put back on the map. Luckily there are two solutions: regional climate models and a process called downscaling. (See following section.) 

 Regional Climate Models (RCMs)

With participation from the Oregon Climate Change Research Institute (OCCRI), CIRC researchers Philip Mote, David Rupp, and Sihan Li completed a unique Regional Climate Model (RCM) project that involved simulating the regional climate of the Western United States with the help of thousands of volunteers who ran the team’s climate simulations on their personal computers. The project, weather@home, employed the Hadley Centre’s regional climate model HadRM3P, and was completed through the distributed computing effort This allowed the researchers to create a superensemble, which is essentially multiple model runs employing different initial conditions (the precise state of the atmosphere and land surface at the start of a simulation) and perturbations of the RCM’s physics, creating what was, in effect, thousands of experiments.



  • First ever regional-scale use of a superensemble using a regional climate model.

  • The use of the weather@home platform allowed the CIRC team to vary the physical processes simulated in the RCM. This allowed for better quantification of physical uncertainties, in this case various small-scale meteorological features, such as the physics of cloud formation.

  • Use of the large ensemble allowed for the detection of 0.1° Celsius (0.18° Fahrenheit)changes in temperature (with 95% confidence) at the scale of 25 x 25 kilometer (15.5 x 15.5 miles) in response to increased C02, compared to ~1 °C (1.8 ° F) from a single climate simulation. This finer detection limit meant spatial patterns of warming could be robustly mapped across the western US.

   Thinking Deeper:

A Note On RCMs & Distributed Computing

Regional climate models (RCMs), as their name suggests, simulate climate for specific regions of the planet.

Because they cover a smaller area than the entire planet, they can provide information at higher spatial resolutions (i.e., finer detail) than global climate models can using the same amount of computational resources. Finer detail alone, however, does not necessarily translate to more certainty in local climate change projections. In fact, the opposite tends to be true. More “noise”—in the form of natural climate variability—frequently accompanies more detail.

From basic statistics we learn that the most straightforward way to find a robust signal in noisy data is to increase our sample size.

In data collection, we can increase our sample size by running an experiment many times. In modeling, we do this by running multiple simulations. Unfortunately, this kind of computing power is often beyond the resources of many research teams. This is where the CIRC team benefited from distributed computing.

Distributed computing allowed our researchers to essentially run multiple model simulations, in effect, running thousands of experiments.


  • By the middle decades of this century (2030– 2049), warming winter and spring temperatures are expected to be greatest at the tops of the Cascades and flanks of the Sierra Nevadas and Rocky Mountains—where mountain snowpack is already disappearing—than at the lower elevations. This is largely due to snow-albedo feedback, but other factors also play a role, such as changes in cloudiness (Mote et al. 2016; Rupp et al. 2017).
  • Each year, roughly 30% of the Pacific Northwest’s winter precipitation falls in heavy, typically atmospheric river–fueled precipitation events.Under future climate change, the Pacific Northwest may experience less warming during major precipitation events, such as those from atmospheric rivers. This could help maintain the amount of precipitation that falls as snow in the Cascade Mountains (Rupp and Li 2017).
  • Physical processes in climate models are represented by equations, some of which may have a parameter in them that must be estimated. For instance, parameter estimates related to precipitation can greatly affect precipitation simulations. The CIRC team’s research using weather@home suggested that the simulation of precipitation can be greatly improved by optimizing the parameter set (Li et al. 2015).



  • Li, Sihan, Philip W. Mote, David E. Rupp, Dean Vickers, Roberto Mera, and Myles Allen.
    “Evaluation of a Regional Climate Modeling Effort for the Western United States Using a Superensemble from weather@ home.”
    Journal of Climate 28, no. 19 (2015): 7470-7488.

  • Mote, Philip W., Myles R. Allen, Richard G. Jones, Sihan Li, Roberto Mera, David E. Rupp, Ahmed Salahuddin, and Dean Vickers.
    “Superensemble Regional Climate Modeling for the Western United States.”
    Bulletin of the American Meteorological Society 97, no. 2 (2016): 203-215. BAMS-D-14-00090.1.

  • Mote, Philip W., David E. Rupp, Sihan Li, Darrin
    J. Sharp, Friederike Otto, Peter F. Uhe, Mu
    Xiao, Dennis P. Lettenmaier, Heidi Cullen, and Myles R. Allen.
    “Perspectives on the Causes of Exceptionally Low 2015 Snowpack in the Western United States.”
    Geophysical Research Letters 43, no. 20 (2016).

  • Rupp, David E., Sihan Li, Philip W. Mote, Karen M. Shell, Neil Massey, Sarah N. Sparrow, David CH Wallom, and Myles R. Allen.
    “Seasonal Spatial Patterns of Projected Anthropogenic Warming
    in Complex Terrain: A Modeling Study of the Western US.”
    Climate Dynamics, no. 48 (2017): 2191-2213.

  • Rupp, David E., and Sihan Li.
    “Less Warming Projected During Heavy Winter Precipitation in the Cascades and Sierra Nevada.”
    International Journal of Climatology 37, no. 10 (2017): 3984-3990.

Multivariate Adaptive Constructed Analogs (MACA) Downscaling Method & Dataset

Global climate models (GCMs)—also called general circulation models—are concerned with the big picture. GCMs are highly complex computer programs that model the whole of Earth’s climate system. They do this by dividing the globe into large three-dimensional, box-like cells. Inside these cells atmospheric processes, such as the formation of clouds, are simulated. GCMs tend to be very coarse in their resolution. This means that they often miss important key local features, including mountains and how they shape climate across the landscape. This is where downscaling comes in.

Downscaling takes the coarse low-resolution data from GCMs and turns it into high-resolution data that accounts for local landscape features and local climate. For CIRC 1.0, University of Idaho researcher John Abatzoglou developed and honed a downscaling method called the Multivariate Adaptive Constructed Analogs, or MACA, method.


  • Advanced the state of the science by refining the MACA statistical downscaling method.

  • MACA data has been used in several CIRC projects, including Integrated Scenarios, Willamette Water 2100, and Piloting Utility Modeling Applications (PUMA) projects.

  • As part of PUMA, MACA data is helping the Pacific Northwest’s two largest water utilities adapt to climate change and variability.


  • MACA demonstrated its accuracy in capturing daily patterns of temperature and precipitation across the complex terrain of the western US (Abatzoglou and Brown, 2012).



  • Abatzoglou, John T., and Timothy J. Brown.
    “A Comparison of Statistical Downscaling Methods Suited for Wildfire Applications.”
    International Journal of Climatology 32, no. 5 (2012): 772-780.

  • Abatzoglou, John T., Renaud Barbero, Jacob W. Wolf, and Zachary A. Holden.
    “Tracking Interannual Streamflow Variability
    with Drought Indices in the US Pacific Northwest.”
    Journal of Hydrometeorology 15, no. 5 (2014): 1900-1912.

  • Abatzoglou, John T., David E. Rupp, and Philip W. Mote.
    “Seasonal Climate Variability and Change in the Pacific Northwest of the United States.”
    Journal of Climate 27, no. 5 (2014): 2125- 2142.

  • Bachelet, Dominique, Timothy J. Sheehan, Ken Ferschweiler, and John T. Abatzoglou.
    “Simulating Vegetation Change, Carbon Cycling, and Fire Over the Western United States Using CMIP5 Climate Projections.”
    In Natural Hazard Uncertainty Assessment: Modeling and Decision Support, Geophysical Monograph 223, First Edition,
    edited by Karin Riley, Peter Webly, and Matthew Thompson. Hoboken: 257-275.
    American Geophysical Union published by John Wiley & Sons, Inc., 2017.
    Print ISBN: 9781119027867.

  • Gergel, Diana R., Bart Nijssen, John T. Abatzoglou, Dennis P. Lettenmaier, and Matt R. Stumbaugh.
    “Effects of ClimateChange on Snowpack and Fire Potential in the Western USA.”
    Climatic Change 141, no. 2 (2017): 287-299.

  • Lute, Abigail. C., John T. Abatzoglou, and Katherine C. Hegewisch.
    “Projected Changes in Snowfall Extremes and Interannual Variability of Snowfall in the Western United States.”
    Water Resources Research 51, no. 2 (2015): 960-972.

Sensitivity-Based Approach to Modeling Watersheds

Rising temperatures and changing precipitation patterns are shifting when and how streamflows occur. But deciphering how climate change might affect any given watershed can be both time-consuming and expensive. Typically the modeling required to see how a given watershed is likely to respond to changes in temperature and precipitation under various climate scenarios is beyond the time, staff, and computational capacity of many.

This is where the work spearheaded by CIRC researchers Julie Vano, Bart Nijssen, and Dennis Lettenmaier comes into play. In several journal articles, these CIRC researchers describe how they applied and improved what’s called a sensitivity-based approach to determine how a given watershed is likely to respond to incremental changes in temperature and precipitation.

The difference in streamflow per degree of warming or precipitation change is the watershed’s sensitivity. The sensitivity analysis honed by Vano, Nijssen, and Lettenmaier is intended as a “short cut” method that can be used in conjunction with more computationally intensive, conventional modeling.

In their several papers on the subject, the researchers not only refined their approach, describing how to apply it to watershed modeling, but also compared how the sensitivity-based approach holds up against more conventional and computationally more costly, “full simulation” modeling.


  • Successfully applied the sensitivity-based approach in the Willamette Water 2100 project.


  • The sensitivity-based approach is comparable to the more computationally expensive full simulation approach in its ability to capture projected seasonality shifts in the hydrologic cycle at the watershed level (Vano et al. 2015).



Modeling Coastal Total Water Levels

A combined threat of rising sea levels, intensifying waves, and major El Niño and La Niña events has led to increased flooding and erosion hazards along the Pacific Northwest coast.As part of our Tillamook County Coastal Futures project, CIRC graduate students Heather Baron and Katherine (Katy) Serafin along with CIRC researcher Peter Ruggiero honed our team’s ability to model this combination of threats using the concept of total water levels (TWLs) achieved at the beach.

TWLs combine factors—including projected sea level rise with increasing wave heights, projected changes in El Niño and La Niña events, the shape of local coastlines, and calculations of the tides—to determine by how much our coastal communities are likely to be inundated by the Pacific Ocean in the decades ahead. For instance, the Pacific Northwest coastline is likely to experience a rise in local sea levels from a couple of inches to roughly 1.5 meters (5 feet) by the year 2100.

However, we can also expect significant differences in the extent of the inundation from location to location due to the elevation of a location’s backshore and any climate adaptation measures adopted by coastal communities, among other factors. What’s more, El Niño events and large tidal events, which temporarily raise local sea levels, can also play major roles. The combination of these factors dictates how high TWL exceeds relevant thresholds for flooding and erosion impacts


Visualizing Total Water Levels 


  • Components of total water level displayed on a sandy, dune-backed beach. (Serafin et al. 2014.)

CIRC Researcher Katy Serafin Describes Her Work

Science in 60: Katy Serafin on extreme coastal events


  • Understanding how total water levels vary from place to place is leading to a better understanding of how sea level rise, changes in storminess, and possible changes in thef requency of major El Niño or La Niña events may impact future coastal flooding and erosion along the Pacific Northwest coastline.

  • Probabilistic total water levels calculations were employed as part of the Tillamook County Coastal Futures project via application of the full simulation total water level (TWL-FSM) model of Serafin and Ruggiero.

  • Using the TWL-FSM approach developed by our team members, we were able to simulate various combinations of events, some of which do not have an analog in the observational record but are physically capable of occurring.


  • Relying only on the observational record may significantly underestimate what areas of the Pacific Northwest coast are at risk of coastal flooding and related hazards (Serafin et al., 2014; Baron et al. 2015).

  • For instance, the 100-year event of extreme Total Water Levels (an event that has a 1% chance of occurring in a given year) could be as much as 90 cm (nearly 3 ft) higher and cause 30% more coastal flooding than previously estimated based on the observational record measured in Oregon’s Tillamook County (Serafin et al. 2014).

  • When Total Water Levels are taken into account, twice as many homes and businesses in Tillamook County would be vulnerable to a 100-year event by the 2050s as are currently considered vulnerable under today’s climate and existing land-use policies (Baron et al. 2015).

  • Changes in wave heights had the most significant influence on Total Water Levels, but sea-level rise had the most impact on shoreline erosion in Tillamook County (Baron et al. 2015).


  • Baron, Heather M., Peter Ruggiero, Nathan J. Wood, Erica L. Harris, Jonathan Allan, Paul D. Komar, and Patrick Corcoran.
    “Incorporating Climate Change and Morphological Uncertainty into Coastal Change Hazard Assessments.”
    Natural Hazards 75, no. 3 (2015): 2081-2102.
  • Moritz, H., Kate White, Ben Gouldby, William Sweet, Peter Ruggiero, Mark Gravens, Patrick O’Brien, Hans Moritz, Thomas Wahl, Norberto Nadal-Caraballo, Will Veatch, 2015.
    “USACE Approach for Future Coastal Climate Conditions.
    Proceedings of the Institution of Civil Engineers – Maritime Engineering, 168, 3, 111-117.
  • Ruggiero, Peter.
    “Is the Intensifying Wave Climate of the US Pacific Northwest Increasing Flooding and Erosion Risk Faster than Sea-Level Rise?.”
    Journal of Waterway, Port, Coastal, and Ocean Engineering 139, no. 2 (2012): 88-97.
  • Ruggiero, Peter, Eva Lipiec, Alexis Mills, John Bolte, Pat Corcoran, John Stevenson,Katherine A. Serafin, and Janan Evans-Wilent. 
    The Tillamook County Coastal Futures Project: Exploring Alternative Scenarios for Tillamook County’s Coastline. Corvallis, Oregon:
    Pacific Northwest Climate Impacts Research Consortium, Oregon State University, 2017.
  • Serafin, Katherine A., Peter Ruggiero, and Hilary F. Stockdon.
    “The Relative Contribution of Waves, Tides, and Nontidal Residuals to Extreme Total Water Levels on US West Coast Sandy Beaches.”
    Geophysical Research Letters 44, no. 4 (2017): 1839-1847.
  • Serafin, Katherine A., and Peter Ruggiero.
    “Simulating Extreme Total Water LevelsUsing a Time-Dependent, Extreme Value Approach.”
    Journal of Geophysical Research: Oceans 119, no. 9 (2014): 6305-6329.