Climatic Research Unit : Research and Projects : STARDEX

STARDEX objectives


Changes in the frequency and intensity of extreme events are likely to have more of an impact on the environment and human activities than changes in mean climate. Losses of life and very high economic damages have, for example, been experienced during recent flooding events in the last two years in Italy, Switzerland, France, the UK and across central Europe. The severe heatwaves which have occurred in the eastern Mediterranean in recent summers illustrate the risks to human health from short-duration temperature extremes. A vital question for Europe is, therefore, whether such events will occur more frequently in the future. For many socio-economic and environmental sectors in Europe there is a clear need for more reliable, high-resolution scenarios of extremes. This is the problem which STARDEX aims to address.

General objectives

Measurable objectives

  1. Development of standard observed and climate model simulated data sets, and a diagnostic software tool for calculating a standard set of extreme event statistics, for use by all partners.
  2. Analysis of recent trends in extremes, and their causes and impacts, over a wide variety of European regions.
  3. Validation of HadCM3 and ECHAM4/OPYC3 climate model integrations, particularly for extremes.
  4. Inter-comparison of improved statistical, dynamical and statistical-dynamical downscaling methods using data from the second half of the 20th century and identification of the more robust methods.
  5. Development of scenarios, particularly for extremes, for the late 21st century using the more robust statistical, dynamical and/or statistical-dynamical downscaling methods.

Specific objectives

  1. To focus on an agreed, standard set of daily temperature extremes (e.g. percentiles of daily maximum/minimum temperature, frost severity and duration indices and a heatwave duration index) and daily precipitation extremes (e.g. maximum length of dry/wet spells, magnitude of the 90th percentile, percentage of rain falling on days with amounts above the 90th percentile) together with derived indices/parameters (e.g. thermal discomfort and fire hazard indices, extreme runoff events).
  2. To focus on specific regions of Europe, selected on the basis of the availability of data and the expertise of the partners, ensuring that the selected case-study regions reflect the range of European climatic regimes and that the size/location of each region is appropriate for the extreme being studied (thus these regions will encompass the British Isles, the Alps, the Mediterranean, Scandinavia and Germany).
  3. To use a consistent approach (in terms of regions, observed and climate model data inputs, variables and statistics studied and time periods) for all analyses and case studies in order to allow rigorous and systematic evaluation and direct inter-comparison of the results.
  4. To analyse observed data series for the second half of the 20th century from specific regions of Europe in order to identify trends in the magnitude and frequency of occurrence of extremes (and, for specific events, their losses in life and financial costs) and to investigate whether these changes are related to changes in other climatic variables (i.e. potential predictor variables derived primarily from NCEP Reanalysis data, such as large-scale and regional objective circulation indices and patterns, including the North Atlantic Oscillation, measures of atmospheric humidity and stability and sea surface temperatures).
  5. To analyse output from the HadCM3 and ECHAM4/OPYC3 GCMs, and RCMs driven both by these two GCMs and by Reanalysis data, focusing on their ability to simulate temperature and precipitation-based extremes (including their magnitude, frequency of occurrence and trends) and potential predictor variables (including their inter-relationships and relationships with surface climate).
  6. To improve existing circulation-based statistical downscaling methods (including methods based on probabilistic weather generators, canonical correlation, multiple regression, neural networks, fuzzy rules and analogue approaches) so that they are able to reproduce observed extremes. This will include the incorporation of additional predictor variables [such as humidity and stability-related parameters (e.g. low-level thermal advection) and sea surface temperatures] in order to address the problem of stationarity (i.e. the underlying assumption of statistical downscaling that observed large-scale/surface climate relationships remain valid under a changed climate).
  7. To calibrate and validate improved ‘regional’ statistical downscaling methods using predictor variables derived from NCEP (1958-2000) Reanalysis data (for calibration and validation under ‘perfect’ conditions, e.g. can past changes in extremes be explained?) and from HadCM3 and ECHAM4/OPYC3 integrations (that include both anthropogenic and natural forcing) for the present day (in order to assess the effects of climate model biases).
  8. To compare the results for specific European regions with output from RCMs (including statistical-dynamical simulations) driven by the same underlying GCMs (i.e. HadCM3 and ECHAM4/OPYC3), with RCMs driven by Reanalysis data, and with results from a two-step analogue approach to statistical downscaling applied European-wide (for a network of 400-500 stations).
  9. To apply the more robust statistical, dynamical and/or statistical-dynamical downscaling methods [identified in 7 and 8 on the basis of (i) present-day validation studies, (ii) inter-comparison of the scenarios obtained by different methods, and (iii) analysis of the ability of the GCMs/RCMs to reproduce the statistics and inter-relationships of the observed predictor variables] to HadCM3 and ECHAM4/OPYC3 (and their associated RCMs) integrations for the end of the 21st century to provide scenarios of extreme events and the associated impacts (such as floods, low-river flow, fire hazard and thermal discomfort) for European regions and to assess the uncertainties associated with these scenarios.
  10. To use these scenarios to identify changes in extremes, to investigate whether these changes are in accordance with recent observed changes and to consider their potential impacts in terms of losses of life and financial costs (based on the impacts of observed changes).
  11. To ensure that the needs of the European climate impacts community for scenarios of extremes are taken into account, that output from the most recent climate model simulations is available for use in the project and that the work is subject to ongoing peer review - by bringing together representatives of the stakeholder (e.g. re-insurance), climate modelling and climate impacts communities in an expert advisory panel.
  12. To ensure wide dissemination of the project results to stakeholders, the scientific community and the public through the project web site and the production of reports, brochures, information sheets and scientific papers.

Last updated: January 2002, Mike Salmon