In this tutorial, I use Uninet, a Bayesian belief net program, to illustrate learning about climate sensitivity from disparate information sources.
I focus on three key concepts:
- Negative learning—how more information can make us less certain
- Deconfliction—how we can deal with conflicting observational signals
- Obsolescence—how seemingly obsolete systems can increase our understanding
The tutorial uses real values from joint research with NASA Langley Research Center.