Spectral identification of plant communities for mapping of semi-natural grasslands

The study was performed on Danish grasslands on well-drained sandy soils. Image data included georeferenced Compact Airborne Spectrographic Imager (casi) data calibrated to apparent surface reflectance. Ecological data included a field-based management map, registration of (vascular) plant species and thirty 30 m by 30 m test sites with affinities seven management classes identified in the field and seven floristic classes modelled from detrended correspondence analysis. Spectral analysis was performed on the derived image reflectance of 18 test sites positioned within the casi scanline. Spectral identification of plant communities was based on a hierarchical approach relating the test sites to i) management (Ma) and ii) flora (Fl) using spectral consistency and separability as the main criteria. Evaluation of spectral consistency was based on unsupervised clustering of test sites of Ma classes 1 to 7 followed by canonical discriminant analysis. Evaluation of spectral separability was based on measures of the Jeffries-Matusita distance. Seed growing generated training classes relating to management and flora (MaFl classes). Maximum likelihood classification showed that the classes were well-defined in statistical terms and also spatially coherent. Superimposition of the classification of MaFl classes on the management map added detailed information of vegetation variation within the management areas. The inverse of the classification accuracy, using the management map as ‘ground truth’, was interpreted as a measure of plant community heterogeneity within management classes. The spectral analysis as well as the maximum likelihood classification indicated that the source of spectral variation within management classes might be related to vegetation composition.