Spectral matching algorithms, such as the Spectral Angle Mapper (SAM), utilize the spectral similarity between individual image pixel spectra and a spectral reference library with known components. Here, we illustrate and quantify the effects that different sources of reference libraries have on SAM classification results. Synthetic images of three mineral endmembers were classified by using reference libraries derived from airborne hyperspectral imagery, ground spectra (Portable Infrared Mineral Analyser), and from a standard library (United States Geologic Survey). Results show that the source of the reference library strongly influences the classification results if all available wavelengths are used. This effect can be partially neutralized by using appropriate preprocessing methods. Two different types of spectral subsetting of the data, two types of continuum removal, and a combination thereof were tested. Best results were achieved by using a feature subset (i.e., limiting the input wavelengths to the diagnostic absorption features). This increased the average classification accuracy from 74% to 95% (ground spectral library) and from 68% to 94% (standard library).

Geologists have been instrumental in shaping Earth observation satellite missions; likewise, geology has been the subject of many remote sensing studies [1]. Applications of optical remote sensing in geology date back to some early studies using the Earth Resources Technology Satellite-1, the predecessor of the Landsat satellite program [2]. In the 1980s, the seventh channel in the short-wave infrared (SWIR) of the Landsat thematic mapper program was added, as a result of spectroscopic mineral studies by geologists [58]. A subsequent satellite-borne instrument, the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), launched in 1999, had specific bands in the SWIR and thermal infrared dedicated to mapping mineral groups [3].

In this paper we present the results of our analyses of multidiurnal low‐resolution Moderate Resolution Imaging Spectroradiometer (MODIS) thermal data for coal fire‐related thermal anomaly detection. Results are presented for data of the Jharia coal mining region of India. We combine three relatively new approaches: first, we use low‐resolution MODIS data for coal fire area analyses, which has only been undertaken by a few authors. Second, we analyse data from four different times of day (morning, afternoon, evening and predawn) and for three different bands (MODIS bands 20, 32 and a ratio thereof); and third, we use an unbiased automated algorithm for thermal anomaly extraction of local thermal anomalies. The MODIS data analysed stem from the years 2001 and 2005. In 2001, MODIS data were only been available from the platform TERRA as morning and evening data (around 1030 and 2200 h). In 2005, MODIS data were available from this platform as well as from the platform AQUA as afternoon and predawn data (around 0130 and 0200 h). Our analyses indicate that MODIS multidiurnal data, and especially bands 20, 32 and ratio bands thereof, have a high potential for the detection of coal fire zones and coal fire hot spot zones, as well as for regular thermal monitoring activities. However, the data are not suitable for a quantitative coal fire analysis concerning fire outline, fire temperature or fire classification into surface and subsurface fires. We used higher‐resolution ASTER and LANDSAT data from 2005 and 2002 for general orientation and later comparison of thermal anomaly extraction results. We also used high‐resolution Quickbird data for the characterization of individual anomalous thermal clusters. Comparisons demonstrate that even low‐resolution thermal sensors such as MODIS can support coal fire detection and zonation into warm and very hot zones.