【摘要】 最佳类比法是基于孢粉数据定量重建古气候的重要方法之一,本文介绍目前在第四纪古气候研究中常用的几种最佳类比法软件和原理的基础上,简要陈述了应用这些软件的数据格式要求、运行步骤和各种方法比较。基于Windows操作系统实现最佳类比的Polygon软件具操作界面友好、易于掌握的特点,但在非相似性距离的阈值和样点类比数目确定方面没有自动判别功能;而使用基于R语言环境的Analogue或MATTOOLS软件的最佳类比法则需要掌握一定的编程基础,但其扩展功能强大,在非相似性阈值和样点类比数目确定上使用ROC曲线(Receiver Operating Characteristic Curve)或者均方根误差(Root Mean Square Error of Prediction)进行判断。相对完善的现代表土孢粉数据库是运用最佳类比法准确定量重建古气候的重要前提,本文尝试R语言环境的Analogue软件包基于东亚现代表土孢粉数据当中的1749个样点数据对湖北神农架大九湖沼泽剖面孢粉数据进行定量古气候重建;根据对现代数据集年平均温度均方根误差的判别结果,选择5个表土样点作为最佳数目进行温度因子计算,获得较好的重建结果。目前的东亚表土孢粉数据库以及大量的现代孢粉分布研究结果为利用类比法进行第四纪古环境重建奠定了基础。更多还原

【Abstract】 Modem Analogue Technique(MAT)is an important approach in quantitative paleoelimate reconstruction based on fossil pollen data.The purpose of this paper is to introduce how to validate software performing analogue calculation,the input data form and running process.The MAT software basing on Windows system are easy and friend to be used,such as Polygon software,which is lack of deciding dissimilarity distance and the number of analogue;but running those software basing on MS-DOS or Linux system need a series of program scripts,such as R-analogue or R-MATTOOLS packages,which can decide dissimilarity distance and the number of analogue using the ROC(Receiver Operating Characteristic)curve or RMSEP(Root Mean Square Error of Prediction).The dissimilarity between fossil pollen spectra and modem pollen assemblage is frequently measured using a number of different metrics,such as Squared Euclidean,Squared Standardized Euclidean and Squared chord distance.Based on fossil pollen data,the modem pollen database is crucial for quantitative paleoclimate reconstruction using MAT approach.The present work used fossil pollen data in DJH core to validate R-analogue software in reconstructing paleoclimate,basing on the threshold of dissimilarity result,the number of analogue is five.The result shows that the East Asia Pollen Database(EAPD)is avail for quaternary paleoclimate reconstruction using MAT approach.更多还原