Stochastic models for climate reconstructions – how wrong is too wrong?

In the last few years, new methods for climate field reconstructions have been proposed. Apart from some stationarity assumptions they usually rely on having an i.i.d. temperature process from year to year. In this contribution we investigate how a mismatch between the used model and the observed process parameters can introduce additional errors into the reconstruction. We conclude that the error introduced by the model mismatch is usually less severe than the one caused by the rather noisy data recording process in the climate archives used.

WERNER Johannes P.;  TORETI Andrea;  LUTERBACHER Juerg; 

2015-11-10

IEICE

JRC90584