@misc{oai:ir.soken.ac.jp:00000779,
 author = {若浦, 雅嗣 and ワカウラ, マサツグ and WAKAURA, Masatsugu},
 month = {2016-02-17, 2016-02-17},
 note = {This study is analysis of anomalies of surface air temperature in Japan. The surface<br />air temperature anomalies relative to the seasoma1 variations are ofour great concern <br />from a long-term forecasting viewpoint. The result of the amalysis presents us the <br />usefu1 knowledge not oniy for the climatic amalysis but also for the prediction and the <br />weather risk management. <br />　In this paper, to begin witza we investigate seasonal periodicities of the time series <br />and show that the surface air tenrperatue has the intense seasonality and there are <br />seasonal periodicities in not only the mean but also the variance. Under the strategy of <br />detection of the yearly distinctive variablities that is the object ofthe predictio". the <br />means and variances ofthe deterministic seasonal periodicities are removed from the<br />original tenrperature data and the residuals are defined as the anomalies. <br />　The low-pass fikered anomalies represent the yearly distinctive5 variablities <br />quantitatively and analysis ofthe monthly divided dataset suggests seasonality in the <br />anomalies. Then a particular parametric form for a nonstationary autoregressive (AR) <br />model is considered to analyze seasonality in the anomalies and the new knowledge is <br />Shown. The model shows that there are seasoma1 changes in the autocorrelation of <br />surface air temperature and the daily power spectrum transformed from the coefficients <br />ofthe model clarifies the seasonal feature. Applying the model to the high-pass filtered <br />datasets clarifies the infiuence of the Japan Current on the seasonality and it is. <br />expected that the long-tenn prediction might be improved by taking the effect from the <br />Japan Current as an exogenous factor. On the other hand, applying the mOdel to the <br />pricing ofthe weather derivatives shows that we cannot neglect the seasonality in the <br />valuation ofthe weather risk in the future. <br />　Furthermore, the model is extended multivariate model. In analysis using the noise <br />contribution, the relation and causality between stations is shown and the structure of <br />the surface air ternperature in Japan is explained. The daily noise contribution <br />estimated from the Multivariate model can quantitatively grasp the clearly seasonal <br />change and the propagation of the causality, and suggests that there are the local <br />teleconnections between stations. The knowledge Will also be import factors' for the <br />prediction., 総研大甲第1041号},
 title = {Analysis of Surface Air Temperature Anomalies},
 year = {}
}