干旱区生态环境知识资源中心

Fuzzy c-means clustering (FCM) is an exploratory data-analysis method that identifies groups of samples with similar compositions. In spite of FCM being well established in the field of pattern recognition, to date it has had little application in biostratigraphy. In contrast to the hard clustering methods commonly used in biostratigraphy, FCM has the advantage that it can accommodate mixtures and! or gradations between clusters. This is an important feature for biostratigraphical data analysis because such data sets often include samples that are transitional between two or more "pure" faunal or floral assemblages. As an evaluation of FCM we used it to resolve Jurassic miospore and pollen biofacies from the Pentland, Fulmar, and Heather formations from two closely spaced wells in the Hawkins Field, Central North Sea, UK. These data were chosen because they contain a flora for which there is substantial paleoecological literature and, as such, would provide a suitable evaluation of the application of the FCM method to industrial biostratigraphical data. The results demonstrated that FCM could extract floral associations that were relatable to stratigraphy and sea-level changes. Fuzzy c-means produced a five-cluster (i.e., five assemblages) solution. We named each assemblage after the taxon that was dominant in the assemblage. The Cyathidites assemblage is stratigraphically distinct and restricted entirely to the nonmarine Pentland Formation. The remaining four assemblages (Perinopollenites elatoides, Lycopodiumsporites, Cerebropollenites mesozoicus, and Callialasporites) are ecologically distinct and occur primarily in the Fulmar and Heather formations. The P. elatoides assemblage is representative of lowstand regression. The C. mesozoicus assemblage is indicative of warmer, drier, possibly semiarid or seasonally arid climatic conditions. The Lycopodiumsporites assemblage is transitional between cooler, wetter to warmer, drier climatic conditions with the floral dominance of P. elatoides being replaced by C. mesozoicus. The Callialasporites assemblage is interpreted as representing warm, wet seasonal climatic conditions, possibly a back-mangrove biotope, with its maximum development occurring slightly above the maximum flooding surface. Ultimately, this succession of curves was used as a proxy for sea-level changes in the study area, enabling the recognition of maximum flooding surfaces, genetic sequences, and parasequences that improved inter-well correlation.