• Semi-supervised linear discriminant analysis

      Toher, Deirdre; Downey, Gerard; Murphy, Thomas Brendan; Science Foundation Ireland; Teagasc (Wiley, 02/07/2012)
      Fisher's linear discriminant analysis is one of the most commonly used and studied classification methods in chemometrics. The method finds a projection of multivariate data into a lower dimensional space so that the groups in the data are well separated. The resulting projected values are subsequently used to classify unlabeled observations into the groups. A semi-supervised version of Fisher's linear discriminant analysis is developed, so that the unlabeled observations are also used in the model fitting procedure. This approach is advantageous when few labeled and many unlabeled observations are available. The semi-supervised linear discriminant analysis method is demonstrated on a number of data sets where it is shown to yield better separation of the groups and improved classification over Fisher's linear discriminant analysis.
    • The significance of the differences in soil phosphorus representation and transport procedures in the SWAT and HSPF models and a comparison of their performance in estimating phosphorus loss from an agriculture catchment in Ireland

      Nasr, Ahmed Elssidig; Bruen, Michael; Moles, Richard; Byrne, Paul; O'Regan, Bernadette (TWRI, 02/07/2012)
      Phosphorus transported from agriculture land has been identified as a major source of water pollution in a large number of Irish catchments. Models of this process are required in order to design and assess management measures. This paper reports on the comparison and assessment of two of the most promising physically-based distributed models, SWAT and HSPF, with particular emphasis on their suitability for Irish conditions. The representation of the overall soil phosphorus cycle is similar in both models but there is a significant difference in the level of detail in describing the chemical and biochemical processes in each model. Also there are differences in modeling the mechanisms by which phosphorus is removed from the soil column and either transported in dissolved form with the runoff water or in particulate form attached to eroded or detached sediment. These differences could have a significant influence on performance when using either of the models to simulate phosphorus loss from any catchment. Both models are applied to estimating the phosphorus concentration at the outlet of the Clarianna catchment in north Tiperrary (Ireland). This catchment is small (23km2) and the landuse is mainly pasture on grey brown podozilic soils. The results of model calibration are presented along with an assessment of the usefulness of the model outputs as a water quality management tool.