Since the SAR, much insight has been provided into fundamental issues concerning the nested regional modelling technique.
Multi-year to multi-decadal simulations must be used for climate change studies to provide meaningful climate statistics, to identify significant systematic model errors and climate changes relative to internal model and observed climate variability, and to allow the atmospheric model to equilibrate with the land surface conditions (e.g., Jones et al., 1997; Machenhauer et al., 1998; Christensen 1999; McGregor et al., 1999; Kato et al., 2001).
The choice of an appropriate domain is not trivial. The influence of the boundary forcing can reduce as region size increases (Jones et al., 1995; Jacob and Podzun, 1997) and may be dominated by the internal model physics for certain variables and seasons (Noguer et al., 1998). This can lead to the RCM solution significantly departing from the driving data, which can make the interpretation of down-scaled regional climate changes more difficult (Jones et al., 1997). The domain size has to be large enough so that relevant local forcings and effects of enhanced resolution are not damped or contaminated by the application of the boundary conditions (Warner et al., 1997). The exact location of the lateral boundaries can influence the sensitivity to internal parameters (Seth and Giorgi, 1998) or may have no significant impact (Bhaskaran et al., 1996). Finally, location of boundaries over areas with significant topography may lead to inconsistencies and noise generation (e.g., Hong and Juang, 1998).
Surface forcing due to land, ocean and sea ice greatly affects regional climate simulation (e.g., Giorgi et al., 1996; Seth and Giorgi, 1998; Wei and Fu, 1998; Christensen, 1999; Pan et al., 1999; Pielke et al., 1999; Rinke and Dethloff, 1999; Chase et al., 2000; Maslanik et al., 2000, Rummukainen et al., 2000). In particular, RCM experiments do not start with equilibrium conditions and therefore the initialisation of surface variables, such as soil moisture and temperature, is important. For example, to reach equilibrium it can require a few seasons for the rooting zone (about 1 m depth) and years for the deep soils (Christensen, 1999).
The choice of RCM resolution can modulate the effects of physical forcings and parametrizations (Giorgi and Marinucci, 1996a; Laprise et al., 1998). The description of the hydrologic cycle generally improves with increasing resolution due to the better topographical representation (Christensen et al., 1998; Leung and Ghan, 1998). Resolving more of the spectrum of atmospheric motions at high resolution improves the representation of cyclonic systems and vertical velocities, but can sometimes worsen aspects of the model climatology (Machenhauer et al., 1998; Kato et al., 1999). Different resolutions may be required to capture relevant forcings in different sub-regions, which can be achieved via multiple one-way nesting (Christensen et al., 1998; McGregor et al., 1999), two-way nesting (Liston et al., 1999) or smoothly varying horizontal grids (Qian and Giorgi, 1999). Only limited studies of the effects of changing vertical resolution have been published (Kato et al., 1999).
RCM model physics configurations are derived either from a pre-existing (and well tested) limited area model system with modifications suitable for climate application (Pielke et al., 1992; Giorgi et al., 1993b,c; Leung and Ghan, 1995, 1998; Copeland et al., 1996; Miller and Kim, 1997; Liston and Pielke 2000; Rummukainen et al., 2000) or are implemented directly from a GCM (McGregor and Walsh, 1993; Jones et al., 1995; Christensen et al., 1996; Laprise et al., 1998). In the first approach, each set of parametrizations is developed and optimised for the respective model resolutions. However, this makes interpreting differences between nested model and driving GCM more difficult, as these will not result only from changes in resolution. Also, the different model physics schemes may result in inconsistencies near the boundaries (Machenhauer et al., 1998; Rummukainen et al., 2000). The second approach maximises compatibility between the models. However, physics schemes developed for coarse resolution GCMs may not be adequate for the high resolutions used in nested regional models and may, at least, require recalibration (Giorgi and Marinucci, 1996a; Laprise et al., 1998; see also Section 10.4). Overall, both strategies have shown performance of similar quality (e.g., IPCC, 1996), and either one may be preferable (Giorgi and Mearns, 1999). In the context of climate change simulations, if there is no resolution dependence, the second approach may be preferable to maximise consistency between RCM and GCM responses to the radiative forcing.
Ocean RCMs have been developed during the last decades for a broad variety of applications. To date, the specific use of these models, in a context similar to the use of nested atmospheric RCMs for climate change studies, is very limited (Kauker, 1998). Although the performance of ocean RCMs has yet to be assessed, it is known that a very high resolution, few tens of kilometres or less, is needed for accurate ocean simulations.
The construction of coupled RCMs is a very recent development. They comprise atmospheric RCMs coupled to other models of climate system components, such as lake, ocean/sea ice, chemistry/aerosol, and land biosphere/hydrology models (Hostetler et al., 1994; Lynch et al., 1995, 1997a,b, 1998; Leung et al., 1996; Bailey et al., 1997; Kim et al., 1998; Qian and Giorgi 1999; Small et al., 1999a,b; Bailey and Lynch, 2000a,b; Mabuchi et al., 2000; Maslanik et al., 2000; Rummukainen et al., 2000; Tsvetsinskaya et al., 2000; Weisse et al., 2000). This promises the development of coupled "regional climate system models".
Other reports in this collection