『Abstract
　Weathering occurs over a wide range of scales. To link features through these scales is a major challenge for interdisciplinary weathering studies. Fractal approach seems to be specially useful for this purpose. We introduce a multistep fractal weathering assessment scheme devoted to extract fractal weathering classifiers from texture analysis of the mineral's image. Our scheme enables to quantitatively estimate the global and local information about the geometry of the weathering pattern. This information is basic to develop geometrical indices of weathering, which can significantly enrich the common qualitative and semiquantitative weathering assessment schemes. To justify the fractal approach, a strong statistical self-similarity has been documented for both the weathering and fresh features of two common silica minerals: quartz and biogenic A-opal (phytolith) over four orders of length scales. The procedure is fast, drastically reduces thresholding bias, promises to be universal, it is valid for genetically different minerals and rock types, scale independent, and specially useful for monitoring the changes in the mineral's roughness during the alteration. Two of the proposed classifiers seem to be potentially useful for direct application in the field and be used by nonspecialist.

Key words: scale invariance; texture; roughness; thresholding; quartz; phytolith』

Introduction
Image fractal analysis
　Fractal imaging and thresholding
Materials and methods
Self-similarity of the weathering patterns and fractal classifiers
　Mineral weathering signature or firmagram (C1)
　Generalized lacunarity (C2)
　Edge detection: Local fractal analysis (LFA)
　　Role of the coefficient h
　　Size of the sliding window and range of grid sizes
　　Definition of the centre for an even window
　Heterogeneity of weathering features distribution (C3)
　Continuity and tortuosity of the patterns (C4)
Conclusions
Acknowledgments
References