You are here

Home

How should I decide what kind of climate projection to use in my planning?

There are several advantages and limitations to the different kinds of projections, so each climate question will have a different set of requirements that must be considered.  Here are some general guidelines for choosing projections based on their design intentions:

GCM Projection Limitations RCM Projection (Dynamically Downscaled) Limitations Statistically Downscaled Projection Limitations
  • coarse spatial resolution - can not give information at the sub-grid scale
  • models may not simulate the proper large-scale circulation dynamics of the atmosphere, which causes variables like temperature and precipitation to be inaccurately represented (certain times of the year may be worse than others)
  • no single model is the best
  • rely on GCMs for simulated conditions outside of the region - if the GCM is wrong the RCM will not be using good information to run its high-resolution simulation
  • some smaller scale features (i.e., lakes) may not be properly represented
  • rely on GCMs for simulating the correct large-scale conditions
  • assume that the statistical relationships between large- and small-scale variables are stationary (do not change over time)
  • some downscaling methods do not allow the variability of variables (i.e., temperature or precipitation) to change
GCM projections are likely best... RCM (Dynamically Downscaled) projections are likely best... Statistically Downscaled projections are likely best...
  • when high spatial resolution is not important
  • if you are interested in studying changes at the continental or global scale
  • if you want information for several decades to centuries into the future
  • when you can NOT confidently say the downscaled projections are superior
  • when studying sub-regional climate changes
  • when local climate processes become important (i.e. lake-effects), but be aware of how well those processes are reprented in the RCM
  • if you need a detailed picture of climate involving several interacting climate variables
  • when a description of the uncertainty (model quality) is important because errors in the way climate processes are simulated can be investigated
  • if you do not need a lengthy time series of data (RCMs are sometimes limited to a few decades because of the amount of computing power they require)
  • if you only want to study small-scale climate in the past or immediate future (since they assume stationarity)
  • if there are no RCM projections available for the time frame or location of interest
  • if the GCMs that were used as the source for large-scale information are not heavily biased
  •  

 

FAQ Tags: 
Climate Projections
Using Climate Information