基于地质记录用时域组合模型预测气候变化趋势的初步研究

A STUDY ON APPLICATION OF TIME DOMAIN COMBINED MODEL TO THE PREDICTION OF CLIMATICTREND BASED ON GEOLOGICAL RECORDS

用时域分析组合模型建立了１００万年来６０°Ｎ６月份太阳辐射量时间序列、宝鸡黄土粒度时间序列、渭南夏季风指数时间序列的动态模型．研究结果表明，时域分析组合模型较好地提取了时间序列的信息，得到的几个显著周期Ｔ＝１３３，１００，８９，４１，２３，１９ｋａ，与地球轨道三要素的变化周期接近．用组合模型拟合实测数据，精度是高的；用其预测未来气候替代性指标时间序列的变化情况，发现未来气候有向干冷方向发展的趋势．时域分析组合模型为研究气候变化趋势提供了一种定量分析、预测的方法．

Chinese loess-paleosoil records were regarded as a proxy of paleoclimatic variations during Quaternary and a time domain combined model was used to predict the long-term climatic trend in the future.For a time series,the time domain combined model could be used to determine the number of the significant harmonics and the corresponding significant period at first. Secondly, the residual error series, obtained through subtracting periodic terms from original time series, Could be Simulated with an ARMA(p, q) process.Through Combining the periodic terms and the ARMA (p, q)model,the prediction model of the time series was established.As an example for prediction test,the prediction model of min-month insolation 60°N for Jun in W/m2 Was established.The results show that time domain combined model to predict the climatic time series is available,functional and precise.Finally,this apparatus was applied to two time series respectively, the Baoji loess grain size ratio(<2μm/>10μm) time series (800-0 ka B. P.) and the Summer Monsoon Index(SMI) time senes(139-4ka B.P.), derived from Chemical Weathering Intensity of Weinan loess section.The two results are similar.They all indicated that the climatic trend will be dry and cold in the future thousands years.In addition,the significant penods,of T=133, 100, 89, 41: 23, 19ka, Which were extracted from the two time series,correspond to the periods of variations in the Earth's orbital geometry.