IEEE Geoscience and Remote Sensing Magazine - March 2021 - 266

2018
[94]

Algorithm-V: visual; F: fixed thresholding; B: bispectral; C: contextual.
Spectral region-V: visible; N: NIR; S: SWIR; M: MWIR; T: TIR; MW: microwave.
Solar condition- N: night; D: day; N/D: night and day.

On/off

Off
-

-
-

-
X

X
-

X
-

-
X

-
X

X
-

X
X

-
-

X
-

X
-

-
-

-
2018
[111]

TM, ETM+, OLI, VIIRS, SAR, ASAR

2018
[86]

AATSR, SLSTR

On/off

On/off
-
-
X
-
X
X
X
-
-
X
-
-
-

Off

2017
[119]

SLSTR

X
-
-
-
X
X
-
-
-
X
-
-
-

On

2016
[88]

MODIS, VIIRS

-
-
X
-
-
-
X
-
-
X
-
-
-

On/off

2015
[89]

OLI

-
-
X
-
-
X
X
-
-
-
X
-
-

On/off

2015
[95]

VIIRS

-
-
X
-
-
X
-
-
-
X
-
-
-

On

2014, 2018
[110], [87]

MODIS

-
-
X
-
X
X
-
-
-
-
X
-
-

On/off

2014
[107]

MODIS

-

-
X

X
-

-
-

-
-

X
-

-
X

X
X

X
X

-
-

-
-

-
X

X
-

-
2014
[78]

OLI

2013
[82]

TM/ETM+

On/off

On/off
-
-
X
-
-
X
X
X
X
X
-
-
-

On/off

2012
[17], [91]

VIIRS

-
-
X
X
-
-
X
-
-
-
-
X
-

On/off

2011
[109]

AATSR, ASAR

X

X
-

-
X

-
-

-
X

X
X

X
-

-
-

-
-

-
X

X
X

-
-

-
-

-
2010
[120]

MODIS

2001, 2007, 2009
[103], [77],
[104], [83]

SEVIRI, MODIS, ATSR

Off

On/off
-
-
X
-
-
-
-
X
X
-
-
-
X

On

1984

OLS

-
-
X
-
-
X
-
-
-
-
-
X
-

On/off

[105]

AVHRR

-

-
-

-
X

X
-

-
X

-
-

X
-

-
X

-
-

X
-

-
X

-
-

-
-

X

1981
[108]

AVHRR

1973, 1978, 1979
[4], [101], [102]

OLS

ON-/OFFSHORE
N/D
D
N
YEAR OF PUBLICATION
REFERENCE

SENSOR

V

F

B

C

V

N

S

M

T

MW

SOLAR CONDITION
SPECTRAL REGION
ALGORITHM

TABLE 2. A SUMMARY OF THE REVIEWED METHODOLOGIES USED FOR GAS FLARE DETECTION (LISTED IN TEMPORAL SEQUENCE).

266

FIXED THRESHOLDING
In the fixed-thresholding category, we consider all the methods that apply a single-radiance
threshold on a specific satellite scene acquired
in one or more bands (from the visible to the
infrared) without any band combination to
discriminate pixels containing gas flares. By
using the IR region, this approach may be considered the origin of more complex methodologies that evolved later.
By simply applying a threshold to AVHRR
MWIR brightness temperature, the authors in
[105] unambiguously identified, in most cases, offshore platforms in the North Sea. Better
performances could be guaranteed using the
single-threshold approach applied to SWIR
signals [17]; at these wavelengths, the small
contribution of the background to nighttime
radiance enables a larger signal-to-noise ratio than in the MWIR [106]. Casadio et al.
[17] found that the drawbacks of using SWIR
data are that other hot targets may be detected and that the SAA generates false alarms.
These confounders were mostly removed by
analyzing the persistence of a detected hot
spot over time. Applying a filter for very high
reflectance values removed the false alarms
from the high-energy particles [17]. When
compared to the DMSP-OLS findings, a lower
sensitivity of the AATSR-based algorithm for
identifying gas flares was recorded (the number of detections were roughly 50-times lower) and attributed to the different spatial and
temporal resolutions of the sensors, together
with the larger DMSP-OLS signal-to-noise ratio. The authors in [91] proposed to improve
the precision in the localization of offshore
platforms by adding the invariance position
observed in radar signals to the SWIR-threshold method.
A combination of multiple thresholds
seems to be mandatory when daytime observations are processed to minimize the solar
contamination. The authors in [78] and [107]
used daytime Landsat imagery acquired by
TM/ETM+ and OLI sensors, respectively. The
methods applied fixed thresholds in series (on
NIR, SWIR, and TIR bands for [78] and on VIS,
NIR, and SWIR bands for [107]) to be sure that
other features (e.g., biomass fires, clouds, bare
land, and sandy surfaces) were not flagged as
gas flares. The latter were selected as persistent
pixels that exceeded the thresholds in all of
the selected bands. Despite their promising
results, these techniques have not yet yielded
operational products, and using such kinds of

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

MARCH 2021
IEEE Geoscience and Remote Sensing Magazine - March 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2021
