Water resources

Regulators and water companies need to be able to undertake assessments of the potential impacts of climate change on water resources over coming decades, e.g. in the assessment of yields in a Water Resources Management Plan (see Chapter 17).

Table 19.2 Indicative sensitivity ranges for climate change (Defra, 2006)




2055-85 2085-2II5

Peak rainfall intensity (preferably for


+ 10%

+2G% +3G%

small catchments)

Peak river flow (preferably for larger

+ iG%



The most common method for climate change impact assessment is to use regionally averaged change factors for the key variables such as rainfall, potential evapotranspiration, runoff and recharge. This approach was used for assessment of climate change impacts on water resources using the UKCIP02 scenarios mentioned earlier. These scenarios are applied using a simple perturbation method whereby the observed series of climatic inputs to a hydrological model are changed proportionally according to the UKCIP02 monthly factors. The approach is deterministic, with uncertainty only reflected in the availability of several different scenarios of future climate.

Arnell (2004) examined the impacts of the UKCIP02 scenarios on river flows by perturbing a baseline daily series of weather data for 1961-90 using change factors derived from the climate scenarios for the 2020s, 2050s and 2080s. The modified data were rainfall and weather variables used to model evaporation. The baseline and perturbed data were used to drive a daily water balance accounting model to examine changes in river flows. Substantial changes were found for mean annual flow and also the Q95 low flow (flow exceeded 95 per cent of the time; see Chapter 11). The analysis was based on one set of emissions scenarios used to drive one family of climate models and one type of hydrological model.

The UKCIP02 analysis expressed change factors as 'block' changes for each month, but had no information about variability at other time scales, e.g. daily or inter-annual. The block change factors were sufficient to investigate changes in mean flows. However, as would be expected in a simulation methodology, variability in the time sequences would tend to increase the variance in the simulated quantities. For example, Arnell (2003) had found that including relative variability within the weather data, on annual or decadal scales, made projected decreases in low flows more extreme. The lack of information about short-term variability in the change factor approach is particularly important for simulation of impacts on flood flows, where both the seasonal climate and short-term rainfall intensity variations affect the distribution of extreme high flows. This is why the high-flow impact simulations discussed above were based directly on time-series outputs from a climate model rather than change factors.

The Environment Agency concluded that the UKCIP02 medium high emissions scenario would lead to a significant impact on average river flows across England and Wales by the 2050s. Winter increases of 10-15 per cent were indicated, but lower flows in most rivers from April to December, with late summer and early autumn seeing decreases of 50-80 per cent in some catchments. Overall, this could mean a drop in annual river flows of up to 15 per cent, the '... biggest challenge we need to overcome to ensure that there is enough water for people and the environment' (Environment Agency, 2008). Geographical patterns in the changes are summarised in Fig. 19.4 for the months of January and July.

This analysis has not yet, at the time of writing, been repeated using the UKCP09 probabilistic regional climate projections. However, Wilby and Harris (2006) have shown how this type of information could be used in a study involving an ensemble of four GCMs, two emissions scenarios, two statistical methods for downscaling GCM outputs, two alternative hydrological models and two distributions of hydrological model parameters to compute probability distribution functions for low flows in the River Thames for the 2080s (Fig. 19.5).

The results underline limitations of climate change impact assessments based on a single climate model or impact model. For example, applying either the HadCM3





Fig. 19.4 Percentage change in mean monthly flow between now and the 2050s using the medium high UKCIP02 scenario (a) January, (b) July. Adapted from Environment Agency (2008).

10 to 15 per cent increase 5 to 10 per cent increase

5 per cent increase to 5 per cent decrease

5 to 10 per cent decrease 10 to 20 per cent decrease 20 to 30 per cent decrease 30 to 50 per cent decrease 50 to 80 per cent decrease

Fig. 19.4 Percentage change in mean monthly flow between now and the 2050s using the medium high UKCIP02 scenario (a) January, (b) July. Adapted from Environment Agency (2008).

or CGCM2 climate model simulations in isolation would yield contrasting river flow scenarios for the Thames basin. Hence there is a need for the water industry and governments to assess multiple uncertainties in the context of water resource planning and flood risk management.

0 0

Post a comment