Stability, Variability and long-term forcing in a monthly-averaged temperature data.

Abstract:

A periodic non-autonomous stochastic model is constructed to approximate the monthly-averaged data spanning decades statistically. The model consists of the deterministic part showing the monthly sensitivity plus a long-term forcing and the stochastic noise part implying short-time processes such as weather. The monthly stability, the magnitude of the noise and the long-term forcing are determined by the statistics deduced from the original data based on the assumption that there exists a clear distinction among time-scales in the data. The generally constructed model is applied to global and hemisphere- averaged surface temperature spanning 133 years. The non-autonomous stochastic model successfully obtains the seasonal statistics similar to those from the data and constructs the long-term forcing, which could be interpreted as an additional heat flux caused by on-going global warming. This result is very similar to that given by global climate models. In particular, the spectral analysis of the long-term forcing shows the characteristics of the flicker noise. It may suggest that the spatial and temporal distribution of the long-term heat flux can be explained by self-organised criticality.