Using a Supercomputer to Keep Catastrophic Flooding From Being a Catastrophe

In the spring of 2019, an extreme flood struck the eastern region of Canada. Rivers ran where cars once drove. Thousands evacuated only to return to waterlogged homes filled with mud. Researchers called the weather event a “100-year flood,” meaning such a disaster has just a one-percent chance of occurring each year. But in a world racked by climate change, this statistic may soon be rendered meaningless. Just two years prior, a “100-year flood” struck the same flood-prone area.

Scientists predict that catastrophic flooding will become more common as climate change continues to warm the planet and disrupt weather patterns. The 2017 and 2019 Canadian floods were spurred by rapid temperature increases and large snowstorms in the spring, two factors thought to be worsened by a warming climate.

Recent major flooding events have caused researchers and politicians alike to recalibrate their tactics: If flooding is to become the norm in some communities, how can the devastation be mitigated before it happens?

The first step is planning, and to do this, you must have a reasonable idea when and where exactly these extreme floods will hit. The Climate Change and Hydrological Extremes Project (ClimEx) attempts to do just that. The project developed highly accurate models that can predict and simulate the course and ferocity of future catastrophic flooding in Quebec and Bavaria.

ClimEx is a cooperative enterprise that synthesizes meteorological research with water management practices. It is a collaboration among supercomputing facilities, universities and government agencies in Quebec, Canada, and Bavaria, Germany, including the Leibniz-Rechenzentrum Supercomputing Centre (LRZ).

More than 50 entities are involved in ClimEx, including university professors, climatologists and representatives of governmental agencies. Each group brings something to the table: the scientists provide the meteorological data, the supercomputing center develops the models and the governmental agents create plans to implement the findings.

The most unique aspect of the project is the method by which it creates the meteorological simulations on powerful Lenovo supercomputer technology, a task tackled by researchers at the Ludwig-Maximilians-Universität (LMU) in Munich in close collaboration with the LRZ.

Supercomputing refers to the processing of incredibly complex and data-heavy problems using thousands of systems in tandem. By harnessing the power of their vast supercomputing resources, the research team of LRZ and LMU was able to create meteorological projections of unprecedented detail and accuracy.

The most unique aspect of the project is the method by which it creates the meteorological simulations on powerful Lenovo supercomputer technology.

To start, ClimEx researchers compiled thousands of historical meteorological data for Quebec and Bavaria, as well as projections for the possible impacts of climate change on each location. This data was input into advanced modeling systems run on one of the SuperMUC systems. The latest installation, called SuperMUC-NG, is built on a cluster of liquid-cooled Lenovo ThinkSystem SD650 servers, powered by Intel® Xeon® Scalable processors.

The simulations run on LRZ’s supercomputers are much more advanced than the rudimentary graphics projected behind meteorologists on television. The ClimEx models allow scientists to predict the timing and path of weather events with greater speed and accuracy, largely due to the simulation’s high spatial resolution and its ability to analyze thousands of data points at one time.

After running around 50 simulations for Bavaria and Quebec, researchers gained a total of 7,500 years of climate data per location. The findings were stark.

Between 2080 and 2099, the models showed temperature increases of up to 8°C during the summer in some parts of Europe, and temperature decreases exceeding 12°C in parts of Québec during winter. The models also predict that Quebec and central Europe will become much drier in the summer, and wetter in the winter.

ClimEx researchers presented their data to representatives of the Bavarian Environment Agency and Waterborne Expertise Center of Quebec during two symposiums, held in 2017 and 2019. Equipped with these findings, the local governments can not only better evaluate the flood risks for their regions, but begin developing methods for mitigating impact, such as building dams and preventing construction in flood-vulnerable areas.

Importantly, the algorithms and technology developed to run these climate simulations will have applications far beyond the locations studied by ClimEx. Other locales can adapt the modeling system to generate climate predictions for their own regions, and assess the likelihood of a variety of extreme weather events, such as heatwaves and wildfires.

Lenovo Neptune® technologies have reduced LRZ’s electrical consumption by up to 30 percent while saving the center an approximate 1.4 million euros annually.

There’s an intriguing irony to all of this. While climate models like the ones developed within the ClimEx project are essential tools in preparing for the devastation of climate change, supercomputing centers require immense amounts of energy -- and capital -- to run. Some supercomputing centers use roughly as much energy annually as it takes to power 13,500 households.

LRZ is taking revolutionary steps to diminish its energy emissions. In 2012, the center partnered with Intel and IBM’s x86 division (later to become Lenovo Data Center Group) to install the first water-cooled x86 cluster at its Munich headquarters. LRZ has installed several iterations of that water-cooled system since then, including the first Lenovo Neptune® branded systems in 2018. Lenovo Neptune® technologies have reduced LRZ’s electrical consumption by up to 30 percent while saving the center an approximate 1.4 million euros annually, or enough electricity to power 3,650 German homes for a year.

Lenovo knows well that to be smarter, you have to be greener. The company has taken steps to address its carbon footprint across all aspects of the business, from production and manufacturing to shipping and logistics.

In the 2017-2018 fiscal year alone, Lenovo reduced its total energy consumption by more than 2.5 percent by implementing energy-efficiency projects across operations and installing renewable energy solutions wherever possible. Where direct energy reductions are not technically or economically feasible, the company purchases renewable energy attribute certificates and carbon offsets to balance its energy usage.

Energy efficiency plays a key role in Lenovo’s development of new products. In light of its manifold efforts, Lenovo was recently recognized as one of the top-100 most sustainable corporations in the world. As climate change continues to warm the planet, forcing humankind to adapt, future innovations will have to consider just that: the future.