摘要：
较之气候水文平均态的缓慢变化,特大洪水引发的灾害对人类社会影响更加显著,而关注极端洪水、认识小概率事件需要更长时间序列.本文通过对太湖钻孔的沉积和磁学参数特征研究,对比太湖文物发掘的历史洪水资料,试图多指标定量重建太湖长序列极端洪水.太湖水则碑对1600～1954A.D.特大洪水记录了15次,通过与现代洪水仪器记录对比和论证,其最低4.03m水位相当于1921～2004A.D.观测太湖的年最高水位80％百分位.太湖钻孔中的沉积粒度和磁化率特征捕捉了水则碑洪水序列中的85％的洪水年,同时补充了水则碑洪水漏失的信号.3次能够被历史文献佐证的洪水沉积信号发生在1766A.D.、1875A.D.和1882A.D.,其洪水水位估计在4.0～4.1m,4.1～4.2m和4.13 ～4.23m.频谱分析显示了沉积洪水指标与水则碑洪水指标具有3个同步的重现期,分别约在90～102年、60～ 62年和42～44年.分析历史洪水与PDO一致性的统计关系,获得估计概率为0.17 ～0.20,肯定了太湖洪水年与PDO存在关联,反映出历史洪水的发生与现代过程相同,受到了太平洋季风环流和夏季降水控制.这些结论为延长洪水时间系列、分析小概率事件、认识极端洪水特征和重现期等提供了重要水文依据.%Extreme large floods have caused disaster much severity for human society, when compared with the equilibrium changes of the climate-hydology. Taihu Lake catchment is influenced by the subtropical monsoon climate with dominant summer-precipitation type, leading that the extreme lake levels during flood years has a close relationship with changes in summer monsoon precipitation not only in modern time but also in the historical periods. In the 20th century, several extreme large floods in 1954A. D. , 1991A. D. and 1999A. D. have been diagnosed to be caused by abnormal summer precipitations that were controlled by the Pacific tropical hurricanes and West Pacific summer monsoon; while Chinese historical chorography has recorded a great number of historical floods in Taihu Lake. However,comparing with the modern instrumental records,the historical floods were very hard to get quantitative information from those text descriptions. It is necessary to built long-term series of flood data to understand those events with the small probabilities. On basis of studies of sedimetology and magnetics from core-based lacustrine sediments, the core taking from the center of Taihu Lake, this paper was attempted to reconstruct the historical sequence of extreme flood of Taihu Lak. Meanwhile, we used multiply evidence of flood events by combining historical flood records from historical literature and cultural relics.rnThrough examining the stone countmark-measured flood level with modern observations of the extreme flood levels in the 20 century, we found that the lowest lake level in the 15 years of extreme floods during 1600 - 1954A. D. had a height at 4. 03m a. s. 1.,equivalent to the 80th percentiles of hydrological-gauged lake levels during 1921 -2004A. D. Flood signals from coarse silt-sand sediments and low-frequency magnetic susceptibility (xlf) captured 85% flood years in the history. The sediment records also revealed some historical flood years that have been missed in the stone countmark records. There were three flood years in the sediment records, which can be identified in historical documents in 1766A. D., 1875A. D. and 1882A. D. and the lake flood levels were estimated at 4. 0 -4. 1m, 4. 1 -4. 2m and 4. 13 -4. 23m respectively. Spectrum analysis using the Fast Fourier Transform for three time series of historical lake levels, sedment granularity and magnetic susceptibility repetitively, produced some synchronorous changes at three peaks of the return periods 90 - 102 years,60 -62 years and 42-44 years.rnTo test if the historical extreme floods with lake level heights and flood occurence have statistical relations with summer monsoon changes, by proxy of the Pacific Decadal Osculation index ( PDO) during the last 400 years, we applied for the conditional test for the two variables that follow Poisson distribution, and obtained the estimated probabilities 0. 17 -0. 20. These estimates accepted the hypotheis that means of the two series samples have common population and the two variables are significant correlated, suggesting that the Taihu historical extreme floods were controlled by summer precipitaions and the Pacific monsoon as the same as the modern flood processes in terms of climatic mechnetics. These conclusions provided important hydrological basis for extending flood sequence to the historical time, recognizing the extreme events with the small probabilities, and analyzing the flood amphititudes and return periods.
