Extreme precipitation events have a large impact on society, as they can cause localized flooding, disruption of infrastructure, or agricultural crop damage. Changes in the frequency or intensity of extreme precipitation in a warming climate are therefore of general interest.

The maximum precipitation rate is limited by the available energy and moisture, and as warmer air can hold more water rainfall intensity can be expected to increase. The Clausius-Clapeyron (CC) equation is a thermodynamic equation that relates the water holding capacity of air to its temperature, and gives an increase in moisture of 6 to 7 percent per degree Celsius. It thus seems reasonable to expect that precipitation, and precipitation extremes, increase with a similar factor and this is called CC scaling.

In a more detailed analysis of convective precipitation, O'Gorman and Schneider (2009) find no basic physical principle for CC scaling and suggest that changes in lapse rate and vertical velocities will cause deviations from CC scaling. Indeed, observations show that the intensity of mid latitude extreme precipitation events increases at almost twice this rate (Lenderink 2008, 2010).

General circulation models do show an increase of heavy precipitation with temperature for daily precipitation extremes. An analysis of the models participating in the IPPC AR4 shows an average scaling of 6 percent per degree global warming, with the majority of the models falling between 4 and 10 percent per degree (Kharin2007). The large uncertainty range suggests that some physical processes associated with precipitation extremes are not well represented in the AR4 models, the most likely one being the parametrized convection required at the coarse spatial resolution. Also, the life cycle of individual storms is not resolved by these daily rainfall statistics. Changes in sub-daily or instantaneous rain rates could be more important for impact studies.

Using a high resolution, non-hydrostatic, model Singleton and Toumi analysed an idealized squall line (Singleton2012). They find a 2 times CC scaling for instantaneous precipitation above 24C, which drops to 1 times CC scaling for longer accumulation periods, likely due to the increase in the squall line propagation velocity. They also find that the total rainfall during the event is not limited by the local moisture content, showing the importance of dynamical processes for precipitation extremes.

We continue their work and investigate the precipitation scaling under more realistic conditions, and for a wider range of events. Using the non-hydrostatic weather model Harmonie at a resolution of 2.5 km, we simulate several rainfall events over the Netherlands. We will consider the following questions:

  • How does the precipitation scale with temperature in a high-resolution, convection resolving, weather model?
  • What does an extreme shower look like under a two degree global climate change?

The picture below shows the total precipitation over the Netherlands for a rainy day in August 2004. The atmospheric moisture and temperature in the model are perturbed (plus and minus 2 degrees) to find their influence on precipitation intensity.


We now look in more detail at the hourly precipitation extremes. Below (to the right) is the probability of exceedence for the hourly intensity for the different experiments (black = unperturbed, red = plus2, blue = min2, green = lapse rate changes). These are calculated with (dotted lines) and without (solid lines) a cut-off of 1mm per hour. Comparing the different percentiles, (left of the figure) we find an increase of approximately 14 percent per degree for this particular event.


A more detailed analysis is underway.