Developing a thermal characteristic index for lithology identification using thermal infrared remote sensing data

详细信息    查看全文

• 作者：Jiali Wei^a ; Xiangnan Liu^a ; ^{liuxncubg@163.com} ; ^{liuxn@cugb.edu.cn} ; Chao Ding^a ; Meiling Liu^a ; Ming Jin^a ; Dongdong Li^b
• 关键词：Lithology identification ; Thermal characteristic index ; Multi-temporal thermal feature space model
• 刊名：Advances in Space Research
• 出版年：2017
• 出版时间：1 January 2017
• 年：2017
• 卷：59
• 期：1
• 页码：74-87
• 全文大小：3196 K
• 卷排序：59

文摘

In remote sensing petrology fields, studies have mainly concentrated on spectroscopy remote sensing research, and methods to identify minerals and rocks are mainly based on the analysis and enhancement of spectral features. Few studies have reported the application of thermodynamics for lithology identification. This paper aims to establish a thermal characteristic index (TCI)(TCI) to explore rock thermal behavior responding to defined environmental systems. The study area is located in the northern Qinghai Province, China, on the northern edge of the Qinghai-Tibet Plateau, where mafic-ultramafic rock, quartz-rich rock, alkali granite rock and carbonate rock are well exposed; the pixel samples of these rocks and vegetation were obtained based on relevant indices and geological maps. The scatter plots of TCITCI indicate that mafic-ultramafic rock and quartz-rich rock can be well extracted from other surface objects when interference from vegetation is lower. On account of the complexity of environmental systems, three periods of TCITCI were used to construct a three-dimensional scatter plot, named the multi-temporal thermal feature space (MTTFS) model. Then, the Bayes discriminant analysis algorithm was applied to the MTTFS model to extract rocks quantitatively. The classification accuracy of mafic-ultramafic rock is more than 75% in both training data and test data, which suggests TCITCI can act as a sensitive indicator to distinguish rocks and the MTTFS model can accurately extract mafic-ultramafic rock from other surface objects. We deduce that the use of thermodynamics is promising in lithology identification when an effective index is constructed and an appropriated model is selected.