A7.2 Future changes in the Urban Heat Island

The regional climate model used in UKCP09 include a scheme which represents the land surface within each 25 km gridbox as a uniform surface, with physical properties determined by parameter values representing the average character of the different land surface types within that gridbox (Cox et al, 1999). However, the surface types are defined using a land-surface dataset at a 1º x 1º latitude-longitude resolution (Wilson and Henderson-Sellers, 1985). At this resolution there is no contribution from urban surface types, so the Met Office RCM does not include any influence of the urban surface on climate. Furthermore, the RCM does not include heat storage during the day and heat release at night by buildings, or HEP as a term in the surface energy balance. Thus the UKCP09 projections will not take into account changes to any of the factors, outlined in Section A7.1, which could change the intensity of the UHI. If none of these factors were to change, or changes were not significant, then the UHI would not change, and it would be reasonable to add UKCP09 projections of temperature change to an observed baseline urban climate to obtain an urban climate of the future.

In applications of the UKCP09 model output, some account of urban effects could be taken by using statistical downscaling techniques calibrated against data which included urban influences. Previous work has shown that the intensity of the UHI is stronger under the low wind speeds, high sunshine, and low humidity conditions typically associated with stagnant high pressure situations (Wilby, 2003b and McGregor et al. 2006). For example, Figure A7.2 shows the strong correlation between the occurrence of anticyclonic weather over Eastern England in summer and the frequency of intense UHI episodes. Assuming that these downscaling relationships hold under future climate conditions, any changes in circulation during the summer (see Annex 6) would have the potential to intensify UHI by a further 0.5°C by the 2020s (Wilby, 2008). Although there are subtle differences in UHI projections downscaled from different GCMs, all point to continued intensification of London’s nocturnal UHI and a greater frequency of intense heat island episodes in summer (see Wilby, 2008). These changes are set against a background of more persistent and intense heatwaves over much of Europe and the USA signalled by other studies (e.g. Meehl and Tebaldi, 2004).

Betts and Best (2004) showed that if the HEP remains unchanged over time, statistical downscaling could be viable. However, if the HEP changes in the future, as is possible under different population and energy consumption patterns, statistical downscaling calibrated against the present-day may no longer be valid. For example, Betts and Best (2004) showed that tripling the HEP from 20 Wm-2 (similar to that of the inner London boroughs) to 60 Wm-2 (the Westminster value) significantly altered the average UHI and increased the frequency of extreme UHI events. Even if the HEP is unchanged, statistical downscaling would have to be performed using predictors drawn from the suite of reliable variables in UKCP09 (including air temperatures, precipitation, relative and specific humidity, cloud cover, short-wave radiation and mean sea level pressure). Low confidence in important predictors such as wind speed, and in joint probabilities with other variables, mean that outputs from UKCP09 are unlikely to support conventional statistical downscaling models based on these data. However, probability distributions of changes in predictors such as mean sea level pressure could be used to perturb baseline pressure data and hence estimate sensitivity of simple indices of the UHI (like the frequency of intense heat island episodes shown in Figure A7.2) to changes in atmospheric circulation alone.

Further development of the HadRM3 regional climate model used in UKCP09 is underway to incorporate an updated land surface scheme which simulates separate surface energy balances for the different land surface types, including urban, within a gridbox. This should allow a more realistic representation of the surface temperature and humidity over each land surface type, including a more realistic response to climate warming. A heat capacity term allows for diurnal heat storage and release over the urban land surface, and an additional HEP term allows for the inclusion of this as an input. All these features have been shown to improve the representation of temperature in urban areas in the model, and should facilitate a more realistic representation of the change in urban temperatures over time in response to changes in urban character and extent.