CHARACTERISTICS OF COMPOSITED AVHRR DATA AND PROBLEMS IN THEIR CLASSIFICATION

Unsupervised classification procedures were applied to a temporal sequ ence of fifteen bi-weekly composited NDVI images produced from AVHRR L AC data. Individual examination of the input images appeared to show s ubstantial contamination due to clouds which persist through the compo siting period. These apparent cloud features dominated the results of the clustering procedures. The composites also include large numbers o f far off-nadir pixels. This causes severe spatial smoothing and produ ces a blurred image appearance. Further combining the data to monthly composites largely eliminated the cloud cover problem, but did not nec essarily reduce the frequency of large view zenith angles. Preprocessi ng of high temporal frequency, low spatial resolution data such as tha t provided by AVHRR and the planned EOS MODIS instrument must more eff ectively remove the effects of clouds, correct for anisotropic scatter ing from the atmosphere and bi-directional reflectance from the surfac e, and should be biased towards the selection of near-nadir measuremen ts. The design and processing procedures for MODIS will reduce problem s associated with atmospheric effects and the geometric distortion of pixels, while enhancing the detection and screening of clouds.