Author(s): Qiuwen Chen; Arthur E. Mynett
Keywords: Fuzzy logic; rule generation; algal bloom modelling; chlorophyll ; a; concentration
Modelling of algal bloom is an ambitious and difficult topic due to the complexity of aquatic ecosystem, insufficient knowledge of the detailed processes and mechanism involved and shortage of high quality data. Owing to the ability to deal with imprecise, uncertain data or ambiguous relationships among data, fuzzy logic (FL) has proved to be a useful and practical method in algal bloom modelling. However, common to any FL modelling approach, the definition of membership functions and inference rules remains difficult. Although, generating rules directly from measurements and observations (rather than consulting the expert) has recently been explored, the procedures turn out to be quite cumbersome. In this paper, a robust FL approach is explored to derive a model directly from measurements, using expert knowledge as a reference only. Its strength lies in the capability to combine partial knowledge on processes with partially available data from observations. The approach was successfully tested in a case for the North Sea to model chlorophyll a (Chl-a) concentration. It had also been introduced to the Dutch pilot study of the European Commission project Harmful Algal Bloom Expert System, which involves 13 institutes and universities from nine EU countries. The objectives of the paper are to illustrate how the robust fuzzy model is developed.