DETECTING CLIMATE SIGNALS - SOME BAYESIAN ASPECTS

A Bayesian approach to detecting forced climate signals in a dataset i s presented. First, the detection algorithm derived is shown to be cap able of uniquely identifying several signals optimally. Other detectio n techniques are shown to be limiting cases. Second, this approach nat urally lends itself to rating models relatively according to their pre dictions. Both the accuracy of the model prediction and the precision of the prediction are accounted for in rating models. In general, comp lex models are less probable than simpler models. Finally, this approa ch to detection is used to detect a signal induced by the solar cycle in the surface temperature record over the past 100 yr. The solar cycl e signal-to-noise ratio is found to be similar to 1 but is probably no t detected. Estimates of the natural variability noise are taken from model prescriptions, each of which is vastly different. The Geophysica l Fluid Dynamics Laboratory models, though, best match the residual te mperature fluctuations after the signals are subtracted. The Bayesian viewpoint emphasizes the need for the estimation of uncertainties asso ciated with model predictions. Without estimates of uncertainties it i s impossible to determine the predictive capabilities of models.