Home > Research > Publications & Outputs > A fuzzy decision tree to predict phosphorus exp...
View graph of relations

A fuzzy decision tree to predict phosphorus export at the catchment scale.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published
<mark>Journal publication date</mark>15/12/2006
<mark>Journal</mark>Journal of Hydrology
Issue number3-4
Volume331
Number of pages11
Pages (from-to)484-494
Publication StatusPublished
<mark>Original language</mark>English

Abstract

Qualitative understanding of the processes controlling phosphorus (P) export from agricultural land has been significantly improved in recent years. Problems remain in predicting P losses despite the requirement of tools providing accurate predictions by legislation such as the EU Water Framework Directive. Decision making, not only in the field of diffuse pollution, often relies on limited data. This study is aiming to predict annual P export from agricultural catchments using a very simple approach that concentrates on the functional behaviour of a catchment. Two simple fuzzy decision trees have been established to predict both total P filtered at 0.45 μm (TP<0.45) and particulate P (TP>0.45) export. The predictions are within range of the P export estimated from measured data using discharge–concentration rating curves. The fuzzy method is capable of identifying the catchments having high P export and reproduces the pattern of P export for wet and dry years, especially for TP<0.45. The predicted fuzzy ranges for TP>0.45 export are wide. The available data indicate that single events have a high importance for TP>0.45 export. We assume that an event-based decision tree might be the appropriate approach to constrain the uncertainties. The proposed methodology is simple. For both trees, a classification is made based on only four input variables using fuzzy rules. The rules do not depend on the estimation of numerous parameters but can easily be adapted once new information becomes available. Therefore, the fuzzy system has a high potential to be used as a decision support tool for policy makers.