 |
| Titel |
Exploring unforced climate variability uncertainty as it applies to climate change detection and model calibration statistics |
| VerfasserIn |
C. E. Forest, B. Sansó |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250024413
|
|
|
|
|
|
| Zusammenfassung |
| Climate change detection and climate model calibration provide two areas in multivariate
statistics where the estimated unforced variability of the climate system is a critical input. In
optimal climate change detection, the unforced variability is required both to estimate the
noise component of the signal-to-noise ratio and also to optimize the multivariate
signal pattern. In climate model calibration, the estimated unforced variability is a
critical component in estimating the Bayesian likelihood function for the model
parameter space given the observations. Thus, at the core of both statistical methods,
we require estimates of the unforced variability in the spatio-temporal patterns
of climate change. These are typically estimated from “long” control simulations
of AOGCMs because the observational records are too short to provide adequate
estimates.
We have recently developed a Bayesian hierarchical statistical model that includes
uncertainty in the estimates of the unforced variability in addition to other components of the
model calibration exercise. Specifically, we seek to separate the different sources of error by
using three sources of information: observational records, control runs and forced runs to
estimate the variability. The variability from each source is handled separately in the
statistical model and estimates for each are provided. As a critical test, we will use data from
seven AOGCMs of the CMIP3 archive to provide estimates of the unforced variability from
multiple models. We will explore the sensitivity of the estimated probability distributions
for the model parameters (effective climate sensitivity, rate of deep-ocean heat
uptake, and the strength of the net aerosol forcing) to using individual AOGCM
control runs and from combining the outputs from multiple AOGCMs. By separating
the results from individual models, we will obtain an estimate of the effects of
structural uncertainty in the AOGCMs on their estimates of the variability and how this
influences the distributions of climate system properties. We have also adapted this
statistical model to the climate change detection problem and initial results will be
presented. |
|
|
|
|
|
|