Niche Modeling

Green man over at ClimateAudit.org noted that the requirements for inclusion of your GCM (global climate model) into IPCC are as follows:

* be full 3D coupled ocean-atmospheric GCMs,
* be documented in the peer reviewed literature,
* have performed a multi-century control run (for stability reasons)and
* have participated in CMIP2 (Second Coupled Model Intercomparison Project).

He observes that there are no actual criteria that show predictive skill. I am glad the IPCC are not designing mobile phone networks or market research software. One of the IPCC models called FOALS (aka Planet Alternating Current) is plotted over at The Blackboard.

It must be a function of science by committee that models are promoted without validation criteria. On this basis, if I extended my long persistent (LTP) random climate model to 3D it would meet the criteria.

Refresh browser to get a new prediction.

This is not a fatuous comment, as one of the most efficient ways to validate models is to generate random outputs with similar statistical characteristics as the models you are testing, and treat the random outputs as a model, running them through exactly the same processing steps as the real models. This provides a benchmark for validity. It gives you insurance that your models are something more than online casinos.

In clinical studies and chemical labs the same role is served by reference standards. Most labs would not operate without periodic testing of their reference standards. These tests are often organized by registration organizations such as NATA. Participation and performance at these trials is a condition of registration. Climate science could use something like this.