June 2, 2014

Climate Modeling Pioneer on Higher Resolution, Higher Impact Change

Nicole Hemsoth

“If you burn one molecule of fossil fuel and put that carbon dioxide into the atmosphere, it will remain there for around 100 years on average,” explained NCAR Senior Scientist, Dr. Warren Washington. “Some remain closer to 500 years. The reason for concern is that if we don’t do something now and we’ve already built in a certain amount of warming since this can’t be immediately turned around without some kind of miracle, then it is clear change can only come gradually.”

The only advantage in this fight, other than action to alter a carbon emissions-fueled culture, is being able to accurately model climate patterns. As Washington explains, this means harnessing the powerful cores scientists have available to mesh meaning out of natural developments like hurricanes and heat waves–and include a constant stream  other data, including the distributions of man-made aerosols and carbon emissions. These high resolution models, both localized and over long periods of time, are the key to policy and decision-making that is based on sheer data.

Dr. Washington is one the few who remembers the time before advanced, high resolution models from the time of his early work at NCAR in the mid-1960s where it took one full day to model one day of weather on an IBM 709. During a recent talk at NERSC as part of their Nobel lecture series, which took place last week, Washington provided the historical “big picture” of weather modeling, discussing in context what the addition of ever-higher degrees of resolution, longer term models of patterns, and the continued tacking on of new sources of climate elements (from aerosols produced from fossil fuels, dust storms, organic matter and sea spray, to name a few) has meant to understanding of climate change and forecasting capability.

As he highlighted in one of his slides, the coupling of ever-higher resolution and new sources, is leading to an incredible amount of new detail and fodder for policy making around climate change impacts and mitigatory measures…


Washington marveled at the breadth of change he’d seen in both the climate—and the tools used to model these alterations. “We’ve now able to drill down to the sub-county level of detail in our models,” he noted, “this allows us to give better treatment of coastal areas and mountains because of the detail.” Far from the days of running one full day to see the next in front, “We can often run on around 700,000 processors for our extreme calculations and at lower resolution, between 2,000-5,000. I think we’re learning how to scale our calculations so they run more efficiently,” he said.

One of the early pioneers in atmospheric models, Washington and his team at NCAR transferred lessons learned from that area over to predictive models that factored in more complex elements, including ocean and sea ice patterns. These are still the basis for a great deal of the work that is done on the large systems at NERSC, NCAR and other weather modeling centers.

Below is the full talk from Dr. Washington, which will kick off this month’s more focused coverage on weather and climate modeling systems.