A Landsat data transformation method which was proposed by the author was applied to extract useful information from data of 769 ground feature classification units of worldwide scope. Three most important characteristicvalues--the general radiance level L, the visible-infrared radiation balance B and the band radiance variation vector(direction and speed) V were calculated. Then the 769 class units were sorted into 106 groups based on their natural characteristics. The means and standard deviations of L, B and V values for all the groups were calculated. The distributions of the 106 groups or the 769 units on the number axes of L, B and V, in the planes of L-B, L-V and B-V, and in the space of L-B-V were investigated. Finally, the typical numerical characteristics of the various ground features are discussed in consideration of their worldwide variations in the present paper. A Landsat data transformation method which was proposed by the author was applied to extract useful information from data of 769 ground feature classification units of the worldwide scope. Three most important characteristic values ​​- the general radiance level L, the visible-infrared radiation balance B and the Then the 769 class units were sorted into 106 groups based on their natural characteristics. The means and standard deviations of L, B and V values ​​for all the groups were calculated. The distributions of the 106 groups or the 769 units on the number of axes of L, B and V, in the planes of LB, LV and BV, and in the space of LBV were investigated. Finally, the typical numerical characteristics of the various ground features are discussed in consideration of their worldwide variations in the present paper.