Coal fire quantification using ASTER, ETM and BIRD satellite instrument data

Coal fires cause severe environmental and economic problems.
Although satellite remote sensing has been used successfully to
detect coal fires, a satellite data based concept that can quantify
the majority of the detected coal fires is still missing. Recently,
the determination of fire radiative energy (FRE) has been
introduced as a new remote sensing tool to quantify forest and
grassland fires. This thesis tests the concept of remotely measured
FRE, with a view to ascertaining its potential applicability to
coal fires. It contains an investigation of a new generation of
satellite instruments, including the operational Enhanced Thematic
Mapper (ETM) instrument, the experimental Bi-spectral InfraRed
Detection (BIRD) satellite sensor and the experimental Advanced
Spaceborne Thermal Emission and Reflection Radiometer (ASTER),
which explores the potential of these sensors to determine coal
fire radiative energy (CFRE). Additionally, based on the results of
this analysis, the thesis presents a new, automated ETM and ASTER
data based algorithm, adapted to quantify coal fires in semi-arid
to arid regions in northern China. Field observations carried out
in September 2002 and 2003 in three coalfields in northern China
(the Wuda, Gulaben and Ruqigou coalfields) demonstrate that coal
fire related, surface anomalies are significantly cooler than
forest and grassland fires. The theoretical investigation of the
ASTER, ETM and BIRD instruments outlines the fact that the thermal
infrared (TIR) or mid infrared (MIR) spectral channels of the
ASTER, ETM and BIRD instrument are particularly effective in
registering these ‘warm spots’, whilst the short wave infrared
(SWIR) spectral range is, however, insufficiently sensitive to be
able to register spectral coal fire radiances. The commonly used
bi-spectral fire quantification method (Dozier, 1981) can be
applied to BIRD data in order to quantify relatively large and / or
hot coal fires. However, existing FRE retrieval approaches fail to
quantify coal fires via ASTER and ETM instrument data. In this
thesis, a new CFRE retrieval method is presented, which links the
fire and background TIR spectral radiances to the CFRE through an
empirical relationship. This newly developed TIR method is applied
to visually detected fire clusters from night-time ASTER data, and
from both day- and night-time ETM data, taken from the three study
coalfields in northern China. The ASTER and ETM CFRE values,
calculated via the TIR method, are compared to CFRE estimates from
BIRD data, calculated via the existing bi-spectral method. Despite
the different spatial resolution and spectral properties of the
ETM, ASTER and BIRD instruments, CFRE computed from ASTER, ETM and
BIRD data show good correlations with one another. However, CFRE
retrievals from daytime data appear to be very undependable to
background temperature variations, while CFRE, estimated from
night-time data, appears to be relatively stable. A comparison
between night-time ETM-derived CFRE and the figures given by local
mining authorities for total coal fire induced, coal loss estimates
in the Wuda coalfield gives a clear indication that the overall
dimension of the coal fire problematic can in fact be approximated
via satellite data CFRE retrievals. It is thus expected that CFRE
derived from night-time satellite data will become a crucial tool
in obtaining reliable, quantitative information for coal fires. A
multi-temporal comparison of CFRE retrievals from night-time BIRD
and ETM data, covering the Ruqigou and Wuda coalfields, indicates
that only major shifts or activity changes in coal fire induced,
surface anomalies can be observed by means of these data. These
results, which could only partially be verified by field
observations, indicate that ETM or BIRD data can be used to monitor
major changes in coal fire related, surface anomalies. These data
however cannot entirely replace detailed field observations,
especially in case of smaller and / or cooler coal fire related,
surface anomalies.