全球变化知识资源中心

A diagram of forcings believed to have influenced the winter 2013/14 UK floods.

Advances in model dynamics⁶⁹ and resolution mean that UK weather forecasts now have levels of accuracy five days ahead that were only possible two days ahead just 25 years ago⁷⁰. It is argued that advances in long-range forecast reliability should also apply to longer-term climate projections⁷¹, subject to the processes being similar in both applications⁷². The UK Met Office uses a common dynamical core for weather forecasting, seasonal prediction and climate change projections. This 'seamless prediction'⁷¹ allows understanding of the causes of those common model biases that often appear within the first few days of simulation, potentially leading to improved process representation and bias removal in subsequent models.

GCMs aim to represent variability in large atmospheric circulation patterns, including the drivers discussed above. High-resolution versions of GCMs, with grid spacings of 20–60 km, are currently being developed. At such resolutions, these models give a better representation of the large-scale drivers, such as storm tracks and embedded depressions, that are key to representing events such as those witnessed in DJF1314. Persistent anticyclonic blocking is typically under-represented in GCMs⁷³, affecting mid-latitude flow with implications for UK rainfall projections, but this may be improved in the Atlantic sector with increases in model resolution or other changes that remove key biases⁷⁴.

Furthermore, RCMs are often nested within GCMs to provide high-resolution information over a region of interest. Such models have typical resolutions of 10–50 km and give an improved representation of daily rainfall extremes. But considerable variation exists between different RCMs' downscaling of the same large-scale features to estimate extreme precipitation⁷⁵. RCMs with kilometre-scale grid spacing (convection-permitting resolution) are also now available. These very high resolutions are too computationally expensive for global operation, but applied regionally they give improved representation of local rainfall processes. Climate change experiments have recently been completed at 1.5-km resolution for the southern UK⁷⁶, showing heavier downpours in the future with global warming. This model gives better representation of various aspects of duration and spatial extent of rainfall⁷⁷ and more realistic hourly extremes⁷⁸ than coarser-resolution climate models. Many improvements are linked to the explicit (non-parameterized) representation of convection⁷⁷. Thus the benefits of such resolutions are expected to be greatest in summer when convective storms are most prevalent. Hence, although summer downpours can only be described accurately at convection-permitting scales, changes in UK winter rainfall seem robust when changing from 12-km to 1.5-km resolution⁷⁶. DJF1314 heavy rainfall was a sequence of frontal storm occurrences and so it remains an active area of research to understand resolution benefits for larger storms of this type.

Very-high-resolution RCMs can provide benefits in terms of assessing the consequences of heavy rainfall, such as in DJF1314, for flooding. Ensembles of RCM models are used to drive flood estimate models⁷⁹, but at typical RCM grid-scale additional spatial downscaling is needed⁸⁰. Very-high-resolution RCMs remove the need for additional spatial disaggregation of outputs and provide realistic fine-scale spatial and temporal rainfall information that can be used directly to drive river-flow models.

Seamless projection for end-to-end attribution⁸¹ of flood risk (combining probability and consequence of flood occurrence) implies routine coupling of climate projections with detailed models of river flow and flood extents, including interaction with flood defences and socio-economic details of properties and businesses. Modelling the potential increase in flood risk for Europe by the 2050s suggests that economic losses could increase from an estimated 4.2 billion euros per year in the 2000–2012 period, to 23.5 billion in the 2050s, under a business-as-usual emissions scenario⁸². Interestingly, the bulk of the increase (about two-thirds) could be due to socio-economic development rather than climate change itself⁸². Hydrological models must represent many factors affecting the transformation of rainfall to river flows, including geology, soils and land-use. For the Thames Basin, a 1-km gridded hydrological model (Grid-to-Grid) forced by several 25-km-grid regional climate models from UKCP09 shows an ensemble-mean increase in flood peaks by the 2080s⁷⁹. But considerable variation exists between the 11 ensemble members, with large spatial variations related to catchment properties. At some locations, some ensemble members show increases beyond the range of natural variability. More recent work has taken a bottom-up approach to the impacts of climate change on flooding in Britain, by using hydrological modelling to establish categories of rainfall-to-flood peak response, linked to catchment properties and underlying climate^{83, 84}. Of 1,120 gauged catchments in England and Wales, 35% are categorized as having an enhanced flood response to rainfall changes⁸⁵. That is, any increase in winter monthly rainfall would generate a proportionately larger increase in flood peaks. Catchments in Britain are most sensitive in autumn and winter⁸⁶ when rain falls onto ground generally more saturated owing to less evaporation of soil moisture.

Changes in land use may adjust surface hydrological characteristics, such as infiltration and soil storage, affecting the timing and magnitude of flood response downstream⁸⁷. The effectiveness of flood prevention by increased upland storage (for example, farm reservoirs, wetlands, floodplains) and changes to crop type therefore requires assessment. For the most extreme rainfall totals, these measures may dampen peak flows but not completely remove flood risk; evidence of land-use management impact on catchments larger than 10 km² remains elusive⁸⁸. Although much of the recent UK flooding was in rural areas, the influence of urbanization requires appraisal, particularly given pressures to build on floodplains. Responses to rainfall extremes in urban areas are faster than for natural surfaces (affecting fluvial and pluvial flood risk)⁸⁹, but can be partially mitigated by detention ponds, soakaways, permeable concrete and provision of vegetated areas.

Damage to exposed European coastal assets from raised sea levels and storm surges could cost tens of billions of euros per year without adaptation⁹⁰. Of the mechanisms involved, the most damaging is often the flood component from short-lived storm surges, where low atmospheric pressure and strong winds add to tidal levels, causing extreme coastal water levels. The great flood of 1953, resulting from a large storm surge event on an already high tidal level, caused considerable loss of life in the UK and Netherlands, and ultimately led to the construction of the Thames Barrier. Changes in extreme sea levels around the world have been dominated by variation in local mean sea level⁹¹. Focusing on the UK coastline, UKCP09 RCM projections⁹² indicate that major changes in storm-driven surge frequency are unlikely over the coming decades. However, a robust projection is that the time-mean sea level, globally and around the UK, is expected to continue to rise over the next century and beyond, increasing the height of extremes, and continuing even after any climate stabilization⁹³. Recent estimates^URL: