There is widespread demand for the interpretation of remotely sensed data into ecologically meaningful classes and in particular, for a greater understanding of the temporal and spatial dynamics of wetland vegetation. This process becomes complex when classification techniques are applied to heterogenous vegetation with no clear boundaries between habitat types. As these fuzzy boundaries, or 'ecotones', may serve as environmental indicators in their own right, the development of techniques to analyse the nature and distribution of these becomes an objective in itself. Airborne RS data will be combined with intensive field spectrometry in order to gain a greater understanding of the effects of species composition and within habitat variation, as well as underlying hydrological regimes, on spectral response. Multitemporal and multispectral datasets are employed to determine the most accurate classification techniques. Quantitative approaches to determine the nature of boundary pixels and transitional zones and the utilisation of change detection techniques with previously acquired imagery, are proposed. Two flights at Insh Marshes are to be used in combination with one from the end of the growing season 2003 to assess phytophenological changes in spectral response. Data acquired at a similar site (Achanalt) will be used to assess the wider applicability of the techniques developed.