A large number of laboratory studies on the spectral characteristics of minerals and rocks show that the most common spectral characteristics (Table 8- 1) of natural minerals in the visible to short-wave infrared spectrum (0.325 ~ 2.5μ m) are due to various forms of iron (Fe3+, Fe2+), water (H2O), hydroxyl (OH-) or carbonate (CO2-).
Table 8- 1 Band Range and Mineral Identification
(According to Taranik J V, 1988)
Because of the relatively wide band range of ETM data, there are some limitations in identifying alteration types. Alterations related to metal mineralization are often divided into iron (such as limonite) and argillaceous (such as carbonation, clayey and chloritization), which can be enhanced by combined transformation of different zones. Fe ~ (3+) and Fe ~ (2+) are rich in iron alteration minerals, which show strong reflection in ETM3, and show different absorption characteristics in ETM 1, ETM2 and ETM4 compared with ETM3. The argillaceous altered minerals are rich in groups such as water (H2O), hydroxyl (OH-) or carbonate (CO2-3), and have a strong absorption band in ETM7 band and a high reflection in ETM5 band, that is, there is a weak spectral contrast between the two bands.
According to the performance of TM spectral bands (Table 8-2), through comparative analysis, principal component analysis is used to extract alteration information, because principal component analysis is a method of multidimensional orthogonal linear transformation to remove the correlation between bands, so the principal components obtained are irrelevant, that is, the information between principal components is not repeated or redundant, and each principal component often represents a unique geological significance. In order to minimize the influence of interference factors in information extraction, autumn phase is mainly used when selecting remote sensing data sources. When extracting alteration information related to metal mineralization, these areas should be properly masked first, and then the characteristic principal component analysis-"crosta method" is adopted, that is, the bands related to specific spectral information are selected as input bands, and those unrelated bands are eliminated, so as to reduce interference factors and highlight the target characteristics, thus enhancing the extraction of alteration information anomalies.
Table 8-2 Main Uses of TM Band