Singular-spectral climate models

In this paper, the singular spectrum analysis method is used to construct data models for the Onega Lake basin: time series of air temperature, precipitation, and runoff of the Shuya, Suna, and Vodla rivers. The model is represented by a trend, seasonality, and noise formed from the components of the singular decomposition of a special matrix of the time series of data. The model is used to describe the variability of the time series variability, reconstruct gaps, and perform a forecast. Models with different numbers of noise components included in the seasonal component are analyzed. It is shown that stable estimates are provided by a model composed of components with large eigenvaIues is proposed to consider the estimates of such a model as local climate data and compare them with the results of global climate modeling. Modeling has confirmed typical regimes of air temperature and runoff fluctuations in the main rivers of the Onega Lake basin.