Assessment of Land Surface Complexity In Relation To Information Capacity and the Fractal Dimension in Different Landform Regions Using Landsat Data

Remote sensing images are highly structured, and contiguous pixels of space domain have strong correlations that contain abundant information on land surface structure features and land surface electromagnetic radiation features. The information capacity model, which is a quality evaluation model based on a multi-dimensional histogram, includes local correlations within different pixels. Thus, the information capacity can illustrate land surface structural information more objectively and effectively than other single-pixel calculation models. Our results reveal that the information capacity value correlates well with the meaningful grey level of remote sensing imagery. This high correlation is related to the complexity of terrestrial surface landscapes. Therefore, information capacity, as applied to geoscience, is introduced in this study to demonstrate the spatial differentiation of information capacity of different landform regions. Generally, the information capacity of a mountain is large and is followed in decreasing order by those of the hills and the plains. Moreover, the correlation between information capacity and the fractal dimension is analysed. Based on the results of this study, it can be concluded that the level of correlation for information capacity and the fractal dimension is high, and the correlation coefficient for the basic landform areas and the loess landform areas is 0.874 and 0.825, respectively. Finally, this paper proposes that information capacity be used as a new reference index for geoscientific analysis in quantitative research on the characteristics of land surface complexity