What got the folks here in the burrow thinking about Real Climate AuditsTM was a report on the Past Millenia Climate Variability Workshop in Eos. They came up with an interesting test of paleoclimate proxy reconstructions.
The PR Challenge will follow a doubleblind protocol in which various international climate modeling centers will provide long-term climate model simulation results for use in the project; a subset (network) will be provided to an independent group of scientists who will produce synthetic proxy data networks with varying statistical characteristics that reflect the behavior of real-world proxy data, with their potential strengths, limitations, and biases. Climate scientists involved in paleoclimate reconstruction will be challenged to apply their methods, using the synthetic proxy networks and the ‘modern instrumental record’ (i.e., actual model climate fields of the nineteenth to twentieth centuries, appropriately degraded to simulate actual instrumental data), to estimate the true past behavior of the model as measured by various data (hemispheric mean temperatures, surface temperature fields, sea level pressure fields, etc.). Further information is available at http://www.pagesigbp.org/science/initiatives/challengeReal Climate AuditsTM are not naive mechanical tests of coding or statistics.
What are some examples of "naive mechanical tests of coding or statistics?"
ReplyDeletewww.climateaudit.com
ReplyDelete