Simulation of extreme weather events present and future

Prof. dr. ir. H.A. Dijkstra, Dr. F. Selten, Dr. G. Burgers, Dr. P.J. van Leeuwen

Institute for Marine & Atmospheric Research [UU] and Royal Dutch Meteorological Institute [KNMI]

From available observational data it has now been well established that the globally averaged surface-air temperature of the Earth has risen by 0.6 °C over the 20^th century. This warming has an impact on many areas in the world. Glaciers have shown a widespread retreat over the last century and globally averaged sea level has risen by 10-20 cm. In the executive summary of the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), it is projected that the globally averaged surface temperature may increase by 1.4 - 5.8 °C by the year 2100.

In recent years, there have been a number of flooding events over Western Europe as well as other extreme weather events, such as extreme winds and hot spells. One of the central questions in climate research is: will the probability distribution of the extreme events become different, from the one in the 20th century, because of a warmer climate? On the basis of the current observational database, however, it cannot be reliably assessed whether or not the probability of extreme events has changed over the last decades. Estimates of very rare events that occur 1 in 1,000 or 10,000 years (for example, important for the design of dikes) are based on the assumption that the available time series is stationary. We know that this it is not the case.

Given the great impact on society of extreme winds, rainfall or hot or cold spells, we propose in this project to calculate the probability of such events using ensemble integrations of a state-of-the-art climate model. To assess any changes in this probability distribution due to global climate change, integrations will be performed for present day atmospheric composition and for a projected change in atmospheric composition (the business as usual scenario from the IPCC) up to the year 2100. These calculations will allow us (i) to validate the methods used to assess very rare events on the basis of short records and (ii) to explore possible changes in extreme events due to changes in atmospheric composition.

Recently, we have ported a state-of-the-art climate model, the Community Climate System Model (CCSM) of the National Center for Atmospheric Research. This model is the next, which will be made available for the researchers to run on several platforms. The model support group of the Centrum voor Klimaatonderzoek, CKO of the has recently suggested that the capabilities of this model should be exploited together with the available computing power in the Netherlands to solve one of the challenging problems in climate research. A problem which is both very challenging and which is also highly relevant to climate changes in the Netherlands is the issue of change in probability distributions of extreme weather events.

To tackle this problem, a multi-member ensemble of runs of 140-year simulation interval will be made with CCSM. This will done for the period 1960-2100 in which the atmospheric composition is changing due to anthropogenic emissions of greenhouse gases, from 2002 onwards.

The ensemble run, with saved daily values of wind, precipitation and temperature for the Atlantic-European area, allows us to calculate the probability of extreme events and changes therein due to changes in atmospheric composition. We think the generated data will be of use for people working in agriculture, water engineering, electric power generation and coastal engineering. Data of the large-scale circulation in the climate model may also be used to drive regional, high-resolution, atmospheric models to better simulate extreme events.

The project will run over approximately 14 months and consume approximately 60 days of compute time on 256 processors of TERAS (=308640 CPU-hours).