REMOTE SENSING - GIS BASED GEOENVIRONMENTAL STUDIES IN JHARIA COALFIELD, INDIA, WITH SPECIAL REFERENCE TO COALMINE FIRES

Abstract

Mining invariably generates a number of significant environmental changes. Jharia Coalfield (JCF), the study area, has a long mining history spanning over a century. Both opencast and underground mining methods are practised in the JCF, as coal is present in multiseams. Several known mine fires (surface and subsurface) occur in this field. Coal fires are a serious hazard. In the JCF, indiscriminate exploitation of coal has led to land-use and environmental changes. The main objective of this scientific pursuit has been to use remote sensing - GIS techniques for land-use studies and coalmine fire studies. Land-use studies have included identification of various land-use classes in the JCF on processed blackand- white multispectral images and on colour enhanced image products, mapping land-use classes, detecting changes in land-use classes in the last two decades, and deriving a correlation between surface thermal anomalies and vegetation anomalies in the JCF. Fire studies have included mapping surface and subsurface fire-areas, temperature estimation of fire-areas, subpixel area and subpixel temperature estimation of some surface fires of limited spatial extent, mapping changes in the fire scenario and depth estimation of subsurface fires. The JCF comprises sediments of Gondwana Super Group of Permo-Carboniferous age. The rocks in the Jharia basin have a general strike of E-W and gentle dips

towards south. The sediments comprise mainly sandstones, shales and a number • of coal seams. The study has been based on the following data: Remote sensing data (including Landsat MSS data of 1975, Landsat TM data of 1990 and 1994, IRS LISS-II data of 1990), Survey of India toposheets, available data and maps pertaining to geology, structure, land-use, subsidence, opencast mining and fire-areas, and selected field observations and measurements. The work has been carried out in a GIS environment using a PC-based ILWIS package at the Department of Earth Sciences, University of Roorkee, Roorkee. Generation of a GIS data-base has involved digitisation of the Survey of India toposheets to serve as the base map. Further, digitisation of all other maps has been followed by rasterisation and interpolation of data, and geoferencing of all the data sets to bring them to a common scale and coordinate system. The main image processing and GIS functions used to process the digital data include contrast manipulation, edge enhancement, image smoothing, statistical normalisation, change detection, density-slicing, colour-coding, generation of various combinations of FCC, IHS transformation, principal component analysis, image ratioing, generation of NDVI images, overlay/logical operations, neighbourhood operations, annotations etc. to extract and highlight features of interest. Interpretations have been made using elements of photo-interpretation and geotechnical elements. The GIS approach has been very useful for land-use mapping. The various land- use classes viz. dense vegetation, sparse vegetation, fire, opencast mining (coal), overburden dump, subsidence and barren wasteland, settlement, transport network, vi ABSTRACT river and water pond, have been identified on the basis of processed black-andwhite multispectral images, colour enhanced image products and field surveys. Image characters of these land-use classes have been given in detail. Integrated interpretation has been made, on the basis which a sample land-use map of a part of the JCF has been prepared. Change detection studies in the JCF have involved comparison of multisensor data (MSS, LISS and TM) of different dates, which have shown very interesting general trends. Further, a detailed analysis has been made on TM 1990 and TM 1994 data of November months. The need for data normalisation for comparative study has been shown. Change detection study has involved image differencing, image ratioing and differencing of ratio images. Colour display of the change detection images has involved thresholding (by density-slicing), colour-coding, and IHS transformation. Change detection images highlight that opencast mining has increased extensively in the JCF; vegetation has undergone a general decrease; however, new plantations have come-up at some places; settlements and transport network have expanded; and fire-areas have changed with time. The thermal anomalies have also been found to be associated with vegetation anomalies, as evidenced from processed satellite data. Surface and subsurface coalmine fires occur in the JCF. Broad-band TM6 data have been found to be useful for detecting, mapping and estimating temperatures of subsurface fire-areas. SWIR bands, TM5 and TM7 have been found to be useful for studying surface fire. Principles of TIR sensing, temperature estimation from remote sensing data and utility of SWIR data for investigating high temperature phenomena have been reviewed.

It has been found that surface thermal anomalies due to subsurface fires have larger spatial extents and relatively lower temperatures as compared to the thermal anomalies due to surface fires. For subsurface fire studies, field data has guided the selection of a threshold DN-value to discriminate non-fire areas from the subsurface fire areas. In the JCF, as inferred from winter time TM6 data of 1990, the ground temperatures associated with subsurface fires range between 25.6°C to 31.6°C. On this data, 231 anomalous TM6 pixels have been detected, which are localised in 55 different sites. Temperature zoning has also been detected within the fire-areas. Pixel-integrated temperatures of surface fires have been estimated to range between 217°C to 410°C. On the TM 1990 data, 59 pixels show anomalous DNvalues associated with surface fires. These pixels are distributed in 17 different sites. In the JCF area, those pixels which are associated with surface fires and show higher than background DN-values in TM5 and TM7 (but no saturation in these bands) have been selected to compute subpixel area and subpixel temperature of the surface fire. For all such computation, a dual-band method has been used, and the results have been derived graphically. The supixel area has been found to range between 0.2 of a pixel (=180m2) to 0.003 of a pixel (=27m2). The subpixel temperature corresponding to these areas has been found to range between 341.7°C to 731 °C. For mapping of subsurface fires, density-slicing colour-coding of TM6 data, followed by IHS transformation (using TM4 image as the intensity image) has been found to be very useful. For mapping suface fires, TM FCC753 has been found to be very informative. Afire map prepared using TM data has been compared with the fire map provided by the field organisations. In general there is a good correspondence between the two. The differences in the two maps have been

attributed to factors such as migration of coal fires, extinguishing of some older fires and nucleation of new fires. For depth estimation a conceptual method based on the principle of linear heat flow in a semi-infinite medium has been designed and tested with numerical experiments. In this method, utilising two or three sets of TIR data of different time instances for the same anomaly area and computing a double ratio, the depth of the buried heat source (coal fire) has been computed. In certain cases the knowledge of initial conditions of the heat source is obviated. Respective sample sets of characteristic curves designed for depth estimation are presented. Sensitivity analysis has been carried out. Further, a simple geometric method has also been used for depth estimation. The thermal anomalies have been used to pinpoint the pixels, vertically below which fires exist. The least horizontal distance between the outcrop (detected on visible and near infra-red bands) and the thermal anomaly has been computed. With the field information on the dip of the strata, depth of subsurface fire has been computed. Limitations of the method have been indicated. The study highlights the importance of Remote sensing - GIS techniques for geoenvironmental studies in mining areas