Online Weather Generator report 3
The Weather Generator perturbed with UK Climate Projections Change Factors reproduces the daily weather variability simulated directly by the Regional Climate Model.
The perturbations (referred to earlier as Change Factors) that will be applied to the WG (in UKCP09) will be taken from the probabilistic projections and joint probabilistic information developed in Chapter 4 and Chapter 5 of Murphy et al. 2009. This information is available for monthly, seasonal and annual time periods, at 25 by 25 km spatial resolution and for each of the seven future 30-year periods (overlapping every 10 years) from 2010–2039 to 2070–2099. The probabilistic projections and joint probabilities give ranges over which the climate (as described by precipitation, mean temperature, daily temperature range, vapour pressure and sunshine) is projected to change in the future (both in the mean and additionally for precipitation and temperature in variance, and for precipitation in skewness, proportion of dry days and lag-1 autocorrelation). As the WG works at the square scale, all squares within the same 25 by 25 km RCM grid box will have the same set of Change Factors. The 5 km grid squares which cross the boundaries of the larger 25 km RCM grid boxes (which are offset from the National Grid based 5 km squares) have been linearly interpolated from the values of the larger grid boxes.
The purpose of this section is to illustrate the way the WG will be perturbed. Observational data are neither long enough to assess whether the proposed way of perturbing the WG will work, nor do they contain a large change in climate when compared to that anticipated in the future. So, instead of using the observations, we use the base RCM integration (the member of the eleven RCMs, available through UKCIP, with the standard set of RCM parameter values — see Murphy et al. 2009 for more discussion on the perturbed physics experiments). For this exercise the WG was re-fitted to the control run period (model years 1961–1990) for selected 25 by 25 km grid boxes across the UK. Change Factors, for this exercise, were calculated from the relevant statistics from the 2070–2099 future and the control-run integrations (see the example later in Table 2). These were then applied to the WG calculated as either differences or ratios depending on the variable (see the Annex for how this is accomplished for the precipitation measures). For mean daily temperature and temperature range, changes were assumed to be the same for all four rainfall transitions within the WG (see the Annex for a discussion of rainfall transitions).
As the WG generates the weather variables in sequence (see Annex), changes to precipitation will affect mean temperature and temperature range, and similarly changes to these variables will affect the generation of sunshine and vapour pressure. We need to ensure that future changes that occur for all non-precipitation variables are exactly the values prescribed from the future RCM integrations. To achieve this we modify the perturbations we apply to these variables to allow for the changes that will have occurred earlier in the generation sequence. This is best illustrated with an example: the selected perturbation for a summer month might be a 50% reduction in rainfall and 3ºC increase in mean temperature. From observations, dry summer months are generally warmer, so there will be a precipitation-related change in temperature of about +0.3ºC. In this example, the change in temperature is adjusted to 2.7ºC so in the generated sequences, the mean change will be equal to the perturbation defined by the Change Factor choice. This procedure becomes more complex for sunshine and vapour pressure, but is essential to ensure that the averages of the generated sequences reproduce the Change Factors developed in Murphy et al. 2009.
Figures 5(a) and 5(b) show plots for the RCM grid box including Heathrow Airport for the future using the perturbation procedure described above. In these plots we have the means and ranges of the 100 generated sequences, together with the crosses (for this one RCM simulation), which is the direct RCM average for the future 30-year period centred on the 2080s. We also show in this figure the differences (in other words the climate change component) compared to the RCM control run. For almost all variables and half months, the direct RCM future values (the crosses) are within the ranges generated by the WG. Table 2 gives the one set of Change Factors used in Figure 5 for the 25 by 25 km grid box encompassing Heathrow Airport.
At this point it is important to realise that for UKCP09 the WG has been fitted using observational data (see earlier and the Annex), then perturbed (with the Change Factors) according to the procedure just described. Using the WG fitted to observational data across the UK will better reflect local topographic and coastal influences on our weather than can be simulated by the RCM. These aspects will remain essentially unchanged in the future, so will be incorporated in the future generated sequences, to the extent that such influences are captured by daily weather data at the 5 km resolution.