Evaluating climate models: Should we use weather or climate observations?

Conference · Tue Dec 01 04:00:00 EST 2009

OSTI ID:982160

Oglesby, Robert J ^[1]; Erickson, III, David J ^[1]

---

1. ORNL

Calling the numerical models that we use for simulations of climate change 'climate models' is a bit of a misnomer. These 'general circulation models' (GCMs, AKA global climate models) and their cousins the 'regional climate models' (RCMs) are actually physically-based weather simulators. That is, these models simulate, either globally or locally, daily weather patterns in response to some change in forcing or boundary condition. These simulated weather patterns are then aggregated into climate statistics, very much as we aggregate observations into 'real climate statistics'. Traditionally, the output of GCMs has been evaluated using climate statistics, as opposed to their ability to simulate realistic daily weather observations. At the coarse global scale this may be a reasonable approach, however, as RCM's downscale to increasingly higher resolutions, the conjunction between weather and climate becomes more problematic. We present results from a series of present-day climate simulations using the WRF ARW for domains that cover North America, much of Latin America, and South Asia. The basic domains are at a 12 km resolution, but several inner domains at 4 km have also been simulated. These include regions of complex topography in Mexico, Colombia, Peru, and Sri Lanka, as well as a region of low topography and fairly homogeneous land surface type (the U.S. Great Plains). Model evaluations are performed using standard climate analyses (e.g., reanalyses; NCDC data) but also using time series of daily station observations. Preliminary results suggest little difference in the assessment of long-term mean quantities, but the variability on seasonal and interannual timescales is better described. Furthermore, the value-added by using daily weather observations as an evaluation tool increases with the model resolution.

• Research Organization: Oak Ridge National Laboratory (ORNL)
• DOE Contract Number: AC05-00OR22725
• OSTI ID: 982160
• Country of Publication: United States
• Language: English

Related Subjects

54 ENVIRONMENTAL SCIENCES
BOUNDARY CONDITIONS
CLIMATE MODELS
CLIMATES
CLIMATIC CHANGE
EVALUATION
RESOLUTION
SIMULATORS
STATISTICS
TOPOGRAPHY
WEATHER