92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Thursday, 26 January 2012: 12:00 AM
The CONVEX Project–Using Observational Evidence and Process Understanding to Improve Predictions of Extreme Rainfall Change [INVITED]
Room 350/351 (New Orleans Convention Center )
Stephen Blenkinsop, Newcastle University, Newcastle upon Tyne, United Kingdom; and S. Chan, E. J. Kendon, and H. J. Fowler

During the last decade, widespread major flood events in the UK and across the rest of Europe have focussed attention on perceived increases in rainfall intensities. Following the summer 2007 floods, which cost the UK over £3 billion, the UK Government increased the annual budget for flood risk management to reach £800 million by 2010. The availability of longer and improved quality observed precipitation series provides more confidence in our estimation of long-term trends and a greater understanding of extreme events within the context of natural variability. There is currently a trend towards increased precipitation and enhanced variability in the high latitudes of the northern Hemisphere, particularly in winter, and observational analyses in Europe suggest significant positive trends in daily rainfall intensities over the past decade. This change is physically consistent with warmer air being able to hold more moisture – General Circulation Models (GCMs) and satellite observations both indicate that the total amount of water in the atmosphere will increase at a rate of 7% per °C of surface warming (the ‘Clausius-Clapeyron' relation) – causing comparable rises in heavy precipitation events, even in regions that may experience a reduction in mean rainfall.

Regional Climate Models (RCMs) are able to simulate the magnitude and spatial pattern of observed daily extreme rainfall events more reliably than GCMs but still underestimate extreme rainfall in relation to observations. Recent research shows that RCM simulation of extreme rainfall is better for long (e.g. 5 to 10 days) rather than short duration events and is best for the winter season when extreme rainfall is mostly associated with synoptic and mesoscale precipitation structures. Particularly during the summer over land a large proportion of precipitation comes from convective storms that are typically too small to be explicitly represented by models. Instead, convection parameterisation schemes are necessary to represent the larger-scale effect of unresolved convective cells. Given the deficiencies in the simulation of extreme rainfall by climate models, even in the current generation of high-resolution RCMs, the CONVEX project (CONVective EXtremes) argues that an integrated approach is needed to provide improvements in seasonal estimates of change in extreme rainfall, particularly for summer convective events. As usable predictions require the bringing together of observations, basic understanding and models a change in focus from traditional validation exercises (comparing modelled and observed extremes) to an understanding and quantification of the causes of model deficiencies in the simulation of extreme rainfall processes on different space and time scales is needed.

CONVEX therefore aims to contribute to the goals of enabling society to respond to global climate change and predicting the regional and local impacts of environmental change on timescales from days to decades. This is addressed through five main objectives: (1) Improved understanding and identification of the spatial-temporal characteristics of extreme rainfall processes using principally UK and European observed precipitation datasets; (2) Assessing the influence of climate and Numerical Weather Prediction (NWP) model parameterisations on the simulation of extreme rainfall; (3) Assessing the influence of model resolution on extreme rainfall process representation by running 1.5km-resolution NWP and climate models; (4) Developing a process understanding of the relationships between large-scale predictors and extreme rainfall on different spatial and temporal scales, including analysis of the first climate change run of a 1.5km RCM, to provide improved estimation of changes to local scale convective rainfall and thus flash floods; (5) Assessing the strengths and limitations of uncertainty estimates derived from ensembles, and the provision of new seasonal estimates of change. This new understanding will be applied in the context of flood risk management.

In addition to outlining the CONVEX project we present results identifying extreme rainfall events using extreme value techniques, such as seasonal maxima and peak-over-threshold analysis. Appropriate diagnostic statistics for hourly and daily precipitation datasets that may be useful in the validation of RCM simulations are presented. Results from the validation of new RCM simulations undertaken by the UK Meteorological Office at 50km and 12km resolutions demonstrate how these may be used to highlight deficiencies in the models. These will subsequently be compared with the 1.5km RCM simulation to identify scale dependencies in the simulation of extreme events. Finally, initial results in identifying process-links between observed large-scale atmospheric circulation and extreme rainfall events using hourly and daily precipitation datasets are considered.

The work undertaken by CONVEX will provide multiple benefits to the modelling and climate impacts communities. It will allow us to understand which extreme rainfall situations benefit from higher resolution which will be used to assess the reliability of coarser model predictions, and if possible, provide quantitative information regarding any deficiencies in the coarser model output. As well as providing improved process-understanding vital for future climate model development and better forecasts from NWP models these results will ultimately provide valuable insight into the characteristics of convective-scale models and into the relationship between models of different resolution that can be applied in the context of climate change predictions. Furthermore, CONVEX will provide new estimates of changes in extreme rainfall to help inform future adaptation strategies for UK flood risk management.

Supplementary URL: