Comparison of SPOT cloud cover quality index against meteorological surface observations

A quality index, based on cloud contamination within the image, is assigned to each SPOT scene. In this study, the information content, of such a quality index, in terms of cloud cover, was tested by comparing it with coincident meteorological surface observation. Global observations during the period June-October 1995 were analysed, starting with the investigation of the spatial and temporal variability of cloud covers from both kinds of observations. Comparisons for a series of 'dependent' datasets generated by varying the time (from 0.5 to 1.5 h) and space (from 5 to 100 km) matching criteria were performed successively. Results were reported as 5 x 9 cloud-class contingency tables and relevant statistics evaluating the correlation between the two observations. The analyses showed an overall good agreement between the two cloud-cover estimations, the correlation was slightly lower (10%) than that obtained by comparing SYNOP against SYNOP. Compared with SYNOP observations, SPOT tended to slightly underestimate cases of broken cloud-cover. One of the most important sources of disagreement was the lack of quantitative information in the three-class cloud quality index code found in 30% of the SPOT images used in this study. When processed as described here, the information contained in the SPOT cloud cover quality index is consistent with the surface observation of cloud cover.