RCMs driving datasets
Regional climate models (RCMs) are what is known as limited-area physical climate models, because their integration domain covers only a sub-region of the globe in order to produce climate simulations at a finer spatial resolution (e.g. 25 km instead of 250 km, as in an Earth system model).
The meteorological conditions at its lateral boundaries must be provided from an external source of global data to allow it to continually keep up with the climate of the rest of the world. This process is called driving an RCM.
Regional climate simulations can be grouped into three main types of equal importance, each with a distinct usage objective, based on the origin of the driving data used at the lateral boundaries:
1. RCM simulations driven by reanalyses
Regional climate simulations driven by reanalyses are reserved for the study of past climates and are not used for climate change studies.
Since they are a representation of the actual global climate, reanalyses provide the best framework for the RCM to attempt to simulate reality. When an RCM is driven by reanalyses, it is expected to accurately simulate most large-scale annual or seasonal characteristics. However, given the chaotic nature of the climate system, it is unrealistic to expect to reproduce the exact sequence of all observed weather phenomena in a regional simulation.
In modelling, this type of simulation is mainly used to validate an RCM in order to improve it, and to deepen understanding of the climate system. It is also used to reconstruct data from the recent past, because an RCM generates consistent series over time and space, with no missing data for over a hundred variables, at a finer resolution than the reanalysis used to drive it. In this case, it offers a greater choice of variables than the various monitoring networks. This is an attractive advantage for users who need data for their own models or who want to develop new models or deepen their understanding of aspects of the past climate related to their field of expertise.
2. RCM simulations driven by a global climate simulation from an Earth system model (ESM) based on observed historical concentrations of greenhouse gases and aerosols
This type of simulation tests the ability of an RCM-ESM pair to reproduce the various climate statistics of a given recent historical period (e.g. 1950-2015), without calibration or use of observations. If this combination of models were perfect, the climate obtained from such a simulation would be similar to one calculated directly from a reanalysis. At least, that’s the goal for climate models. However, even though GHGs and aerosols are very real, the timeline of simulated weather events from an ESM is completely virtual and has nothing to do with the timeline observed in reality. As a result, the simulated and observed time series will not be the same, but the different statistical moments of their distribution should be similar. It is never possible to compare a particular event, month, season, or year in a climate simulation from an ESM-driven RCM with the observed data. Only statistics calculated over several years of simulation can be compared with statistics calculated over several years of observations in the same period. This comparison between simulated and observed statistics is only possible because the RCM-ESM pair used the observed GHG and aerosol concentrations.
These historical regional climate simulations driven by ESMs are used as a reference in climate change studies carried out by both users and modelling teams.
3. RCM simulations driven by an ESM climate projection up to the end of the 21st century and even beyond
A climate projection by an ESM is based on future scenarios of GHG emissions or concentrations and aerosols. A climate projection produced by an RCM-ESM pair shows how the regional climate might change in the future in response to some forcing by the GHGs and aerosols. The difference between the results from simulation types 2 and 3 is used to estimate the climate change projected by an RCM-ESM pair. It’s also possible to calculate climate change by comparing two future periods, for example 2071-2100 vs. 2041-2070, but only if they are based on the same GHG scenario. Note that in order to identify a climate projection, it is essential to specify the GHG scenario used and the period covered, as well as the climate models used (for example, CRCM5 climate projection driven by CanESM5 member 1 based on the SSP3-7.0 GHG scenario for the period 2041-2070). This is because the further along we are in the 21st century, the more future GHG scenarios diverge from each other. This nomenclature may seem cumbersome, but it’s very important for all users to comply with it in order to accurately identify all the climate projections used in their work.
