Innovatives Supercomputing in Deutschland
inSiDE • Vol. 10 No. 1 • Spring 2012
current edition
about inSiDE
index  index prev  prev next  next

UPSCALE: Global high-resolution Climate Simulations

Climate-related extreme events, such as tropical cyclones and high-latitude storms, and their associated risks are a major concern for society. Climate scientists currently have limited under- standing of extreme event behaviour and how the nature of extreme weather may respond to climate change.

The UPSCALE project (UK on PRACE: weather-resolving Simulations of Climate for global Environmental risk) aims to address these concerns through computer-intensive simulations of weather and climate. This has been made possible through generous access to the brand new Hermit supercomputing resource in Stuttgart, Germany, access that was granted by the Partnership for Advanced Computing in Europe (PRACE). High-resolution simulations of the global climate system will be used to improve understanding of weather and climate risk, and explore possible future weathers and their risks.

The Joint Weather and Climate Research Programme (JWCRP) team of researchers working on the UPSCALE project comprises climate scientists from the UK Met Office and the National Centre for Atmospheric Science (NCAS) Climate. The team is led by Pier Luigi Vidale, Willis Professor of Climate System Science and Climate Hazards at the University of Reading's Meteorology Department and Director of the Weather and Climate Hazards Laboratory. The UPSCALE project capitalizes on the joint climate modelling expertise brought together by the JWCRP partners. During the project the UPSCALE team will conduct a series of experiments on the Hermit supercomputer. Each experi- ment will dynamically simulate 25 years of the climate. One ensemble of experiments will simulate the current climate, in order to gain deeper understanding into the impact of natural climate variability on the behaviour of climate extremes. A second experiment ensemble will simulate future climate conditions using future climate scenarios, allowing the team to explore how extreme weathers may change in the future. This is particularly important for end-users of extreme weather information, such as decision-makers in the insurance industry, the travel sector and utility companies.

The climate model used for these experi- ments is the UK Met Office Hadley Centre global climate model HadGEM3 at 25km horizontal resolution. The Hadley Centre Model, ran at such high- resolution, is essentially the same model as used for the Met Office's global weather forecasts. The ensemble approach, running the model multiple times for each climate state, provides a larger source of extreme event simulations leading to more robust results.

The increased resolution is crucial for representing fundamental weather and climate processes more completely. Past studies have shown that high- resolution allows small-scale weather processes to be resolved, improving both the simulation of high-impact events, such as tropical cyclones and European blocking and associated European summer and winter extremes. Addition- ally these small-scale features play a vital role in the overall mean climate and its variability, particularly at regional and local scales, which in turn improves the simulation of extreme weather event activity. Tropical cyclones, for example, are relatively small-scale events occurring over relatively short time periods. However, their location, frequency and intensity are influenced by large-scale, often global scale, climate variability operating on temporal scales from weeks to multiple decades. Improved understanding of current and future weathers and their risks therefore challenges climate modeling and computational capabilities.

The experimental design therefore targets climate model resolution, simulation length and experiment sample size simul- taneously, which is necessary for the sufficient simulation and understanding of climate-related risk.

Access to Hermit presents the opportunity to run the global climate model at very high resolution, for extended periods of multiple decades, which has not been possible with the available national computing resources in the UK. In fact the proposed experiments would take 33 years to complete using the existing resource allocation, rather than within the 12 months project time.

Using a global approach to exploring extreme event occurrence opens up a further opportunity. Global climate mod- elling allows the scientists to investigate potentially connected high-impact events, such as the 2005 hurricane season and coinciding drought in the Amazon, or the 2010 Russian heatwave and the Pakistan floods. There is evidence that these concurrent events were related via global teleconnections, which are recurring and persistent patterns of large-scale climate anomalies that can influence the behaviour of extreme weather. An example of a tele- connection pattern is the El Niņo Southern Oscillation (ENSO). Again, high- resolution is vital for simulating the key processes associated with climate teleconnections.

Using experience gained from running the 25km global climate model, the team then aim to achieve simulations at resolutions up to 12km, which is not even envisioned for global weather forecasting before 2015.

• Pier Luigi Vidale
University of Reading

top  top