An examination of the consequences which radiometric calibration, atmospheric correction and solar angle correction cause in the analysis of a multitemporal data set.

摘要

Multitemporal, multisensor image analysis relies upon data calibration and atmospheric correction for enhanced accuracy of derived physical units such as radiance and reflectance. This experiment subjects a multitemporal data set to: radiometric calibration, atmospheric correction, and solar elevation angle correction. The effects of these corrections per band, on vegetation indices, and potential effects on supervised classification are described. The results show those data in Landsat MSS bands respond differently as a function of correction. Band 6 data, unlike the other three, is relatively unaffected by the corrections. Vegetation indices (VIN, NDVI) are strongly dependent on each data correction. Radiometric and atmospheric correction both increase the index and solar elevation correction reduces it. Discriminant analysis reveals that the Mahalanobis distance between cover class groups and the effectiveness of the classification functions are both increased by the corrections in a summer time data set.