Jackson School of Geosciences researchers use Lonestar to improve land models and predict the regional effects of climate change: There is no doubt that climate change will affect our lives. Rising temperatures will alter agriculture, sea levels, and coastal ecologies and increase the number and strength of extreme weather events like hurricanes and floods. But how these effects will be felt, where they will occur, and how soon, is unclear. Dealing with the impact of climate change on the environment is one of the greatest challenges of the 21st century. To preempt a catastrophe, scientists around the world are simulating the future changes on powerful supercomputers, making climate modeling one of high-performance computing’s (HPC) most important responsibilities.
For decades, HPC has helped scientists simulate the global environment and make the case for climate change. And now, more than ever, supercomputing centers like the Texas Advanced Computing Center (TACC) are playing a crucial role in predicting how climate change will alter the atmosphere, oceans, ice-caps and land-surfaces, and demonstrating how human action, or lack thereof, can hasten or slow the pace of change.
A Prediction Is Only as Good as its Model
Climate modeling involves four major components: air, ocean, sea ice and land. Compared to relatively homogenous systems like oceans and air, land-surface processes are more diverse, complex, and difficult to represent.
Drs. Zong-Liang Yang and Guo-Yue Niu, experts in land-surface and climate modeling at The University of Texas at Austin’s Jackson School of Geosciences, and their team of climate researchers, use the Lonestar supercomputer at TACC to create and test new models that quantify climate change and assess the impact of those changes on Texas.
“It’s very challenging because of the heterogeneity of the land surface. You need to consider many detailed processes, down to the micropores in the soil and its impacts on the evaporation to the atmosphere,” Niu said.
Leaf transpiration, crop irrigation, stream flow. All the complicated environmental interactions that occur on the land need to be transformed into accurate, coupled equations and added to the model. “These are very small scale processes that need to be considered on the global scale,” Niu said. “That’s the major challenge.”
(a) GRDC (Global Runoff Data Center) runoff climatology (observation), (b) the modeled runoff, and (c) the modeled water table depth. These simulations are the first to model global water table (or groundwater level) dynamics, and the first time to represent groundwater dynamics in Global Climate Models (GCMs), which are powerful tools for projecting climate change.
For more than 15 years, Niu has been developing land surface models and contributing his insights to the National Center for Atmospheric Research (NCAR) Community Land Model, a primary tool used to predict the consequences of future climate change on human environments. Niu’s current project on Lonestar uses advanced computational methods to parameterize the impacts of various environmental processes and adds new features, like ground water and stream-flow, to the model for more realistic simulations and more accurate predictions.
“We mainly work on model development based on an understanding of the fundamental processes and then representing those processes using equations,” Niu said. “We’re interested in many scientific problems, but we don’t trust the tools that exist now, that’s why we spend a lot of time improving them.”
Land and Water
One of the major limitations of the traditional land-surface model (called Noah) is the lack of lateral dimensionality. To address this issue, Yongsheng Xu, a postdoctoral research associate at the Jackson School, is testing a new 3-D version of the land hydrology model, called Noah-D.
The model collects high-resolution land data, meteorological forcing and rainfall data from remote sensing measurements. It combines this information with new algorithms to upgrade the current 1-D model to a 3-D version that accurately predicts how water will flow and the run-off route it will take to the oceans.
Driving Noah-d with six meteorological variables and NEXRAD Radar precipitation, Xu compares and calibrates Noah-d simulation results with real river flow measurements from gauge stations to improve the model performance. He will also couple the Noah-d output with estuary models to study the impact of the extreme weather events and climate variability on estuary ecosystems.
“The reason that environmental scientists are interested in the land-processes is very simple: humans live in the land,” Xu said. “For example, hurricanes form over oceans, but what people are most concerned about are the consequences on land.”
Predicting land processes is critically important to society, but the calculations involved require incredibly powerful computing systems. “Noah-d is a computationally-intensive modeling system. If you use a single processor computer, it takes several months or years to finish a long term Noah-d simulation, depending on the size of the domain.” Xu said. “On Lonestar, with parallel computing techniques, I can use hundreds of processors at once to save significant time. This is the advantage of massively parallel supercomputers.”
The simulation of continental water dynamics for the whole U.S. is a ‘petascale’ problem, according to Xu, requiring computers that can perform trillions of operations per second, which Ranger approaches. “In heavy rainfall events, will it cause a flood? And if there’s a flood, where will it occur? With the right models and powerful supercomputers, we can answer these questions.”
“Ozone Action Day, Ride Free”
Whereas Niu and Xu describe interactions between land-surfaces and hydrology, Xiaoyan Jiang, a PhD student of Yang, works on the science that connects what’s occurring on land to what’s happening in the air.
A model showing simulated ozone concentration over the U.S. on Jun 5-6, 2002. Jiang ran the coupled regional model on Lonestar and generated this animation [click to watch].
Jiang studies the impact of global warming and human development on near-surface ozone levels in Houston, Texas. Houston already has ozone levels well above the national average, and Jiang showed that anthropogenic (“human-based”) factors, like buildings, factories, and refineries, are a significant cause of rising ozone levels in the Houston area.
“When the eight-hour average ozone concentration near the surface reaches more than 85 parts per billion, it’s harmful to people,” she said. “As the temperature goes up and people put more emissions into the atmosphere, ozone levels will go up as well. By knowing how the climate or land use change can affect the ozone levels, people might think about doing something to reduce this harmful effect.”
Her research couples meteorological predictions with chemical reactions models, and uses real air quality data to ascertain how rising levels of greenhouse gases will alter the air we breathe years and decades down the road.
“There are more than 100 chemical reactions in the model,” Jiang explained. “We provide the emission sources, for instance nitrogen-oxides and volatile organic compounds, and the model performs chemical reactions, while calculating the temperature, humidity, wind speed and other meteorological variables.”
Predictably, Jiang found that global warming will exacerbate the ozone and air quality problems in Houston unless something is done.
“How will future climate change affect Texas? Some people have done work for the whole U.S., but there’s no specific work for Texas,” Jiang said. “As climate changes, sea levels will rise and many ecosystems will change. It’s a big problem.”
Conclusion
The climate models that researchers working at the Jackson School of Geosciences develop, and the simulations they produce on TACC’s powerful supercomputers, can only predict our future — they cannot change it.
However, as the models become more realistic and the predictions grow more dire, computational simulations speak forcefully about the need for drastic change. Providing policy makers and the public with information about water management, pollutant emission, and land use, computational modeling alters the debate on how, where and when to act to cope with climate change.
************************************************************************
To learn more about climate modeling at the Jackson School of Geosciences, visit the
Land, Environment, and Atmospheric Dynamics website.
Aaron Dubrow
Texas Advanced Computing Center
Science and Technology Writer
