Written evidence from University of Oxford (FFP 106)

 

Floods occur usually after extreme events like the rainfall associated with storm Desmond in early December 2015 (van Oldenborgh et al., 2016) or in very wet seasons like the record breaking wet January in 2014 (Schaller et al., 2016). Whether or not such events lead to flooding depends on many factors unrelated to the meteorological event itself. The subject of this evidence, however, is our understanding of the extreme precipitation events relevant to pluvial and fluvial flooding.

We focus on the implications of the use of a constant multiplicative “uplift” on observed extreme rainfall to account for the implications of climate change. A necessary starting point for the use of such an uplift is an understanding of what it means in the context of today’s climate. There are two potential issues with the uplift procedure. First, if the uplift is applied to an observed precipitation maximum or some other (ideally more robust) statistic inferred from the observed rainfall record, then it will be subject to statistical uncertainty due to incomplete sampling of the distribution of rainfall to date. Second, the meaning of the uplift in terms of changing return-times depends on the properties of the underlying distribution and how this is responding to climate change. The first issue is a sampling problem that can be quantified with standard statistics: this is not our focus here. We address the second issue: given that extreme rainfall is not expected to increase by a constant factor in all locations and on all timescales in response to climate change, what does the application of a constant multiplicative uplift factor mean?

Using a very large ensemble of simulations of possible rainfall in a single season (DJF 2013/14) obtained from a 50km-resolution regional climate model (Schaller et al., 2016) nested in a global atmospheric model, we first map the return-time corresponding to an event 30% larger than an event with a 1-in-100-year return-time in the ensemble. Figure 1 shows the results for daily and monthly precipitation (panels a and b respectively). Several points are immediately evident: there is considerable spatial variation, and the variation depends on the timescale considered. In the monthly data in particular, a pattern emerges that a 30% uplift on the 1-in-100-year event corresponds to a <1000-year event in the Midlands, but a 2—3,000-year event in the South East. This pattern will almost certainly be model-specific, but it illustrates the potential importance of explicit simulation to quantify what these uplifts mean in the context of any climate, changing or otherwise.

Comparing panels a and b, we see that a 30% uplift on 1-in-100-year daily rainfall corresponds to a markedly less extreme event than a 30% uplift on 1-in-100-year monthly rainfall. This timescale-dependence is particularly important, because UK river catchments respond on intermediate timescales between daily and monthly. If a uniform uplift is applied, it will mean very different things in different catchments, depending on the timescale of the catchment response.

a     b

Figure 1 return times of an extreme rainfall event with a magnitude of 30% above the magnitude of a 1-in-100-year event based on (a) daily and (b) monthly precipitation from a 18,336-member ensemble simulation of the DJF 2013/14 season using a 50-km resolution regional climate model nested in a global atmospheric model. Note the different colour scales in panels a and b.

We illustrate this point in more detail by considering two exemplary gridboxes corresponding to “London” and a location in the Midlands. Figure 2 shows return-times, with uncertainties, for daily and monthly precipitation at these two locations. Solid lines show the magnitude of 1-in-100-year events, while dashed lines show a 30% uplift.

a   Macintosh HD:private:var:folders:9y:wzpx9s7j767_b3s5x0g28m800000gn:T:TemporaryItems:returntimeforgridboxes_monthly_errorbarsandlines_redblue.pdf

Figure 3 return times of (a) daily and (b) monthly DJF precipitation for two exemplary gridboxes. The solid horizontal lines represent the magnitude of a 1-in-100-year event and the dashed lines represent an event of a magnitude 30% above the 1-in-100-year event.

In these particular gridboxes, applying a 30% uplift to 1-in-100-year daily precipitation provides an approximately 1-in-300-year event (only London shown and a location in the highlands to illustrate a very different climate), but applying the same uplift to monthly precipitation yields a 1-in-600-year event in the Midlands, but a 1-in-1100-year event in London.

Since rainfall is not expected to increase by a constant fraction at all locations and on all timescales, the application of a uniform uplift factor will have a different impact on flooding return-times depending on location and the response timescales of the river catchment. The results presented here are based solely on a preliminary analysis of an out-dated version of the Met Office Unified Model (albeit the only version with which it is possible to run these very large ensembles necessary to obtain robust statistics on very rare events). Hence they are not intended to provide quantitative evidence, but to illustrate the potential importance of explicit simulation of the distribution of flood precursor events and how this distribution is changing, rather than reliance on a simple uniform uplift factor. At the very least, if an uplift factor is to be applied, it needs to be clear what return-time event in the current climate it is being applied to, and the implications for return-times in a future climate need to be quantified. It may also be necessary to apply different uplift factors to precipitation extremes on different timescales in order to yield consistent levels of protection across river catchments.

Recent advances in the science of extreme event attribution (NAS report) may provide the necessary tools to provide quantitative estimates of return times on different timescales and for different regions now and in a future climate. Calculating these return times explicitly using independent methodologies as developed e.g., by the Met Office (Christides et al., 2013), the University of Oxford (Pall et al., 2011) and the Royal Netherlands Meteorological Institute (KNMI, van Oldenborgh, 2007), might provide the basis to calculate these uplift factors and assess the uncertainty associated with them. Thus providing valuable information to decision-makers faced with tough questions about changing risks and to underpin resilience strategies at a more local level.

March 2016

References

Christidis, N., et al. (2012), A new HadGEM3-A based system for attribution of weather and climate-related extreme events, J. Clim., doi:10.1175/JCLI-D-12-00169.1

National Academies of Sciences, Engineering, and Medicine. 2016. Attribution of Extreme Weather Events in the Context of Climate Change. Washington, DC: The National Academies Press. doi: 10.17226/21852.

Pall, P., Aina, T., Stone, D.A., Stott, P.A., Nozawa, T., Hilberts, A.G.J., Lohmann, D. and Allen, M.R. (2011) Anthropogenic greenhouse gas contribution to flood risk in England and Wales in autumn 2000. Nature, 470: 382-385

Schaller, N., Kay, A.L., Lamb, R., Massey, N.R., van Oldenborgh, G.J., Otto, F.E.L., Sparrow, S.N., Vautard, R., Yiou, P., Ashpole, I., Bowery, A., Crooks, S.M., Haustein, K., Huntingford, C., Ingram, W.J., Jones, R.G., Legg, T., Miller, J., Skeggs, J., Wallom, D., Weisheimer, A., Wilson, S., Stott, P.A. and Allen, M.R. (2016) Human influence on climate in the 2014 southern England winter floods and their impacts. Nature Climate Change.

van Oldenborgh, G. J. (2007): How unusual was autumn 2006 in Europe?, Clim. Past, 3, 659-668, doi:10.5194/cp-3-659-2007.

van Oldenborgh, G.J., Otto, F.E.L., Haustein, K., and Cullen, H. (2015) Climate Change increases the probability of heavy rains like those of storm Desmond in the UK - an event attribution study in near-real time. Hydrology and Earth System Sciences, 12(12): 13197-13216.