Multitemporal classification of Texas AVHRR imagery using harmonic components

Multitemporal approaches using sequential data acquired over multiple years are essential for satisfactory discrimination between many land cover classes whose signatures exhibit seasonal trends. The seasonal variability can represented by a harmonic model which is characterized by three components: frequency, phase and amplitude. The trigonometric components of the harmonic function inherently contain temporal information about changes of land use. Using the estimates which are obtained from sequential images through spectral analysis, seasonal periodicity can be incorporated into multitemporal classification. The Normalized Difference Vegetation Index (NDVI) was computed for two-day composites of the Advanced Very High Resolution Radiometer (AVHRR) imagery over Texas from 1991 to 1992. Vegetation types were then classified both with the estimated harmonic components and the average NDVI using an unsupervised segmentation approach based on a hierarchical clustering algorithm that incorporates spatial textural information. Results are compared to output from the classification of a typical image observed in each season during the two year period