Coal fire mapping from satellite thermal IR data – A case example in Jharia Coalfield, Jharkhand, India

https://doi.org/10.1016/j.isprsjprs.2005.12.002Get rights and content

Abstract

Coal fire has been found to be a serious problem worldwide in coal mining processes. Coal fire burns valuable coal reserve and prevents access to proven reserve in the affected area. Moreover, it leads to severe environmental degradation of the region by an overall increase of the ambient atmospheric temperature, by the emission of obnoxious gases (e.g., SO2, NO, CO, CH4) along fissures and cracks and by causing land subsidence and collapse. Jharia Coalfield, Jharkhand, India, is known for being the exclusive storehouse of prime coking coal as well as for hosting the maximum number of known coal fires among all the coalfields in the country.

In this paper, some of the important issues of coal fire mapping from satellite thermal IR data have been addressed in particular reference to Jharia Coalfield. Namely, these are: retrieval of true spectral radiance from raw digital data using scene-specific calibration coefficients of the detectors, thermal emissivity of surface materials to obtain kinetic temperature at each ground resolution cell of satellite data, field-based modelling of pixel-integrated temperature for differentiating surface and subsurface fire pixels in Landsat TM thermal IR data, identification of surface coal fire locations from reflected IR data and lateral propagation of coal fire.

Introduction

The temperature of the earth materials and high temperature phenomena can be estimated based on the thermal radiation from these materials and phenomena as described in the Planck's radiation equation (Rees, 2001). The equation relates spectral radiance and wavelength with radiant temperature of an object as follows:Lλ=2hc2λ5hceλkTrad1where,

    Lλ

    Spectral radiance (W/m2/sr/μm)

    λ

    Wavelength (m)

    Trad

    Radiant temperature of the object (K)

    h

    Planck's constant = 6.26 × 10 34 J s

    c

    Speed of light = 3 × 108 m/s

    k

    Boltzmann constant = 1.381 × 10 23 J/K

Eq. (1) can be rearranged as follows:Trad=hc/kλln[2hc2λ5Lλ+1]

In this equation, wavelength may be considered as the mean wavelength of the spectral region under investigation. Given the spectral radiance, obtained from the calibration parameters, radiant temperature of a pixel can be calculated using the above equation. When spectral emissivity (ελ) of a pixel is available, kinetic temperature of the said pixel may be obtained using the following mathematical relation:Tkin=1ελ1/4Trad

Any material on the earth's surface is capable of emitting thermal radiation as it always bears a temperature greater than absolute zero. Nevertheless, it is important that the received thermal radiation energy is discriminated from the background noise. For this purpose, post-calibration cut-off spectral radiances for zero and maximum DN values need to be fixed. Markham and Barker (1986) listed post-calibration dynamic range i.e., minimum and maximum spectral radiance values (LMINλ and LMAXλ) corresponding to the DN values 0 and 255 for U.S. processed Landsat TM digital data acquired during two specific periods: (i) prior to 15 January 1984 (for Landsat 4 data) and (ii) after 15 January 1984 (for Landsat 4 and Landsat 5 data). For determining spectral radiance from DN values, for the purpose of calculating surface temperature and delineating coal fire areas, many previous workers (Prakash et al., 1995, Prakash et al., 1997, Saraf et al., 1995, Zhang et al., 1997) have used these cut-off values for LMINλ and LMAXλ. Surface temperature calculated by this procedure was found to be quite low, which is difficult to consider for the fire pixels particularly in cases where surface fire occupies reasonably wider areal extent of the pixels. Because of high gain setting, the thermal sensor TM Band 6 gets saturated (reaches maximum DN value 255 of 8-bit data) at approximately 70 °C (or 343 K). As a result, surface coal fire when occurring for wide areal extent (e.g. full pixel area) gives pixel-integrated temperature of about 70 °C only, which is actually much higher as observed from field measurements (> 150 °C). In this paper, an attempt has been made to gain new insight in thermal IR data processing and analysis with particular reference to surface and subsurface coal fire mapping in Jharia Coalfield, Jharkhand, India. Some of the important issues in coal fire mapping have been addressed, viz. scene-specific calibration of digital data as a function of the detectors' sensitivity, thermal emissivity of surface materials to obtain kinetic temperature, field-based modelling of threshold pixel-integrated temperature for delineating surface coal fire pixels in Landsat TM thermal IR data, identification of surface coal fire locations from short-wave infrared reflected IR band (TM band 7), and net lateral propagation of coal fire during the observation period.

The Jharia Coalfield is located near Dhanbad town which is 260 km north-west from Kolkata (Calcutta) city and 1150 km south-east from Delhi city in India (Fig. 1a). The major coal-bearing formation in Jharia Coalfield includes the Barakar Formation of Early Permian age forming a sickle-shaped outline in the northern part of the coalfield and the Raniganj Formation of Late Permian age in the southern part (Fig. 1b).

Section snippets

Retrieval of spectral radiance from raw digital data using scene-specific calibration parameters

The relationship between spectral radiance and digital numbers of raw satellite data is called calibration. Pre-launch calibration parameters of the detectors rarely remain unchanged when actual imaging is performed due to a number of reasons. Markham and Barker (1986) listed post-calibration cut-off spectral radiance values (LMINλ and LMAXλ) for minimum (0) and maximum (255) DN values of 8-bit data. However, the sensitivity of the detectors in each band varies among themselves and deteriorates

Data used

Two sets of nighttime Landsat-5 TM data acquired on March 09, 1992 (N-08/200 and N-08/201 scenes covering southern and northern parts of the study area respectively) and April 21, 1996 (A008-200) were used in this work.

During daytime, solar heating shares a significant portion of the total radiant energy flux from the earth's surface. Besides, topographic unevenness may cause differential solar heating. A height difference of about 180 m (ranging from 140 m to 320 m) over the study area may not

Results and discussion

From the nighttime Landsat-5 TM raw data, initially the spectral radiance image has been generated using the average offset digital numbers and gain coefficients of the individual scenes. Subsequently, radiant temperature has been calculated from the spectral radiance values using Planck's equation (Fig. 3a). In this work, attempts have been made to carry out temperature anomaly mapping based on kinetic temperature of the surface materials. For this purpose, thermal emissivity map has been

Conclusions

In this work, scene-specific average calibration coefficients in each band have been used to obtain spectral radiance values for individual pixels instead of using the listed post-calibration dynamic ranges in spectral radiance, LMINλ and LMAXλ, corresponding to minimum and maximum digital numbers in each band (Markham and Barker, 1986). The radiant temperature of the pixels, obtained from these spectral radiance values, is found to be comparable with the average field temperature of the

Acknowledgements

The author gratefully acknowledges the help and cooperation from the staff and officials of Bharat Coking Coal Limited (Coal India Limited), Dhanbad, India. The author is deeply indebted to Dr. V.K. Dadhwal, Dean, Indian Institute of Remote Sensing (IIRS), National Remote Sensing Agency (NRSA), Dehradun, India, and Dr. P.S. Roy, formerly Dean, IIRS, NRSA, Dehradun, India, and presently Deputy Director, NRSA, Hyderabad, India, for their views and suggestions in numerous technical discussions,

References (28)

  • S.C. Banerjee

    Prevention and Combating Mine Fires

    (2000)
  • J.A. Barsi et al.

    Landsat TM and ETM+ thermal band calibration

    Canadian Journal of Remote Sensing

    (2003)
  • Bhattacharya, A., Reddy, C.S.S., 1992. Airborne scanner survey and data analysis for underground and surface coal mine...
  • K.J.K. Buettner et al.

    The determination of infrared emissivity of terrestrial surfaces

    Journal of Geophysical Research

    (1965)
  • D. Chandra

    Jharia Coalfields

    (1992)
  • J.A. Davies et al.

    Field determination of surface emissivity and temperature for Lake Ontario

    Journal of Applied Meteorology

    (1971)
  • C.S. Fox

    The Jharia Coalfield

    (1930)
  • F.S. Kneizys et al.

    Atmospheric Transmittance/Radiance: Computer Code LOWTRAN-7. AFGL-TR-88-0177

    (1988)
  • S.B. Mansor et al.

    Monitoring of underground coal fires using thermal infrared data

    International Journal of Remote Sensing

    (1994)
  • B.L. Markham et al.

    Landsat MSS and TM post calibration dynamic ranges, exoatmospheric reflectances and at-satellite temperature

  • W.E. Marlatt

    Remote and in-situ temperature measurements of land water surfaces

    Journal of Applied Meteorology

    (1967)
  • M. Matson et al.

    Identification of subresolution high temperature sources using a thermal IR sensor

    Photogrammetric Engineering and Remote Sensing

    (1981)
  • A. Prakash et al.

    Surface fires in Jharia Coalfield, India – their distribution and estimation of area and temperature from TM data

    International Journal of Remote Sensing

    (1999)
  • A. Prakash et al.

    Surface thermal anomalies associated with underground fires in Jharia Coal Mines, India

    International Journal of Remote Sensing.

    (1995)
  • Cited by (0)

    View full text