Regional Climate Modelling
Downscaling is the term used by which climate data is used for fine spatial scales to devise Regional Climate Modelling (RCM) Downscaling methods include
- Statistical - making use of the statistically derived relationship between the variable of interest and the larger-scale field.E.G. If a wind field has a high correlation to precipitation then changes in that wind field may infer changes to future precipitation. This requires a lot of information and the correlation may not hold true under changed climatic conditions. So this is better under part or present studies rather than future ones
- Dynamic - The models used for this purpose are usually high resolution global atmosphere models or regional (limited area) climate models. Since the output from these models is dynamically self-consistent and does not involve the use of statistical relationships that may no longer hold under an altered climate regime, this method enables more reliable future projections of climate change to be made at finer spatial scales. It also provides regionally complete information over the region that is being simulated.
- Statistical Dynamical - combines the two approaches outlined above, and provides information at finer spatial scales by using output from a higher resolution model, but exploiting the statistically derived relationships between the variable of interest and the larger-scale field
Regional Climate Model
An RCM can usually be used with a gridbox size of as little as 25 km or even less. RCMs need to be supplied with the values of required fields at their boundaries - these can be taken from global climate models or from analyses of observational data, and these are interpolated to the resolution of the RCM in a ‘buffer’ layer at the edges of the RCM. A GCM, or observations, are used to drive the RCM.
If a regional climate model is to be used to make projections about future climate, it must be driven at the boundaries by output from a global climate model. Since an RCM needs to be supplied with such data, it is often run nested within a global climate model that can provide a continuous supply of boundary data. The information supply is most often one-way, from the global to the regional model.
Realism of regional climate model output
Where terrain is flat we could be confident that the coarse resolution of a GCM might be sufficient but inland their could be mountains and other types of terrain and so the RCM will provide more realistic preidictions.
Ability of regional climate models to represent smaller islands
Another advantage of regional climate models is their ability to represent smaller islands. Both GCM and RCM are reliable but RCM gives a much higher degree of accuracy
Increased detail of regional climate model output
This time projected changes in winter precipitation between the present day and the 2080s are shown for much of Europe. The RCM clearly predicts climate change in much more detail than the GCM. for many mountains and even mountain ranges, such changes will not be seen at all in a global model, but the finer resolution of the RCM will resolve them.
Simulation of extreme weather events
RCMs are much better than GCMs at simulating weather extremes. At all the thresholds, the simulated probability from the GCM agrees less well with observations than that from the RCM, but the agreement also becomes steadily worse as the higher thresholds are approached. At the very highest threshold, namely in the case of extreme wet weather events, it is clear that only the RCM can reproduce observations. Extreme weather events tend to arise from weather systems that are generally smaller in size, so the increased resolution of the RCM is needed to resolve them
Simulation of cyclones and hurricanes
It is not known how the frequency of hurricanes will change as global warming progresses, although there are indications that they will become more severe. Hurricanes are of a size that cannot be adequately resolved by a GCM; the finer resolution of an RCM is essential for this. In the diagram below there is a critical difference; the RCM resolves a tropical cyclone-like vortex in the Mozambique Channel that is not present in the GCM.
Use of regional climate model output to drive other types of model
As said above an RCM is driven by the output from a global model, or by observations. In turn, the output from an RCM can be used to drive other types of model. In particular, the higher resolution data, and associated smaller-scale features, of regional climate models can often be very usefully used as input to other types of model. The diagram below illustrates a tropical cyclone-like vortex in the Bay of Bengal (left panel) simulated by an RCM, the output from which was used as input to a coastal shelf model to produce a map of high water levels in the Bay.
The most comprehensive impacts studies will use an ensemble of RCM outputs to span the possible range of likely behaviour of the climate system in order to make projections about the range of likely impacts on human and natural systems.
Conclusion
Whilst global climate models still have a lot to tell us about the climate system and its sensitivity as a whole, regional climate models will almost certainly be necessary in our attempts to project future changes in climate at the local scale. In addition, ensembles of regional climate model projections will be required in attempting to quantify the uncertainty associated with any such climate forecast.
<comments />