You are on page 1of 11

Preliminary detection efficiency and

false alarm rate assessment of the


Geostationary Lightning Mapper on
the GOES-16 satellite

Monte Bateman
Douglas Mach

Monte Bateman, Douglas Mach, “Preliminary detection efficiency and false alarm rate assessment
of the Geostationary Lightning Mapper on the GOES-16 satellite,” J. Appl. Remote Sens. 14(3),
032406 (2020), doi: 10.1117/1.JRS.14.032406

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Preliminary detection efficiency and false alarm rate
assessment of the Geostationary Lightning Mapper
on the GOES-16 satellite

Monte Bateman* and Douglas Mach


Universities Space Research Association, Huntsville, Alabama, United States

Abstract. The Geostationary Lightning Mapper (GLM) is a new geostationary lightning detec-
tion and location instrument, developed for the current generation of Geostationary Operational
Environmental Satellite-R Series (GOES-R, S, T, U) satellites. Detection efficiency (DE) and
false alarm rate (FAR) are evaluations of what an instrument detects and what it misses, com-
pared with other well-characterized measurements. These are critical parameters to be assessed
for the GLM. We now have enough data to produce preliminary DE and FAR assessments of the
prelaunch predicted DE and FAR performance specifications for GLM. However, one of the
challenges of doing this assessment is that all available ground truth data sets have much smaller
coverage, and in some places, lower DE than GLM. In an attempt to overcome these limitations,
we have created a “virtual” lightning network by clustering flashes from several different
ground-based networks, using a technique similar to the clustering of GLM pixels. All ground
truth sources that are within 330 ms and 16.5 km of any other ground truth source are added to
the current cluster. Treating this as a single, “virtual” network, we calculated the relative DE and
FAR of GLM. We found that the GLM easily meets the DE spec of better than 70% averaged
across the whole field-of-view, over 24 h. It meets the FAR spec in most of CONUS, and we
suspect it will improve with updated processing software. © 2020 Society of Photo-Optical
Instrumentation Engineers (SPIE) [DOI: 10.1117/1.JRS.14.032406]
Keywords: Geostationary Lightning Mapper; GOES-16; detection efficiency; false alarm rate.
Paper 190586SS received Jul. 31, 2019; accepted for publication Mar. 6, 2020; published online
Apr. 17, 2020.

1 Introduction
The Geostationary Lightning Mapper (GLM) is a new, geostationary lightning detection and
location instrument, developed for the current generation of Geostationary Operational
Environmental Satellite-R Series (GOES-R, S, T, U) satellites.1–3 The GOES-R satellite, the first
in the series, is now in orbit and designated the operational GOES-East (GOES-16) satellite.
Here, we detail the detection efficiency (DE) and false alarm rate (FAR) analysis for the
GLM instrument on GOES-16. The challenge of validating the GLM is that it has a much larger
field-of-view (FOV), and at times and places, a higher DE, than most ground truth lightning
detection networks. In order to create the best ground truth dataset possible for comparing
to GLM, we combined several ground-based lightning detection networks into a single “virtual”
network database.

2 Instrumentation
The GLM is a staring, CCD imager aboard GOES-16.1 The GLM optically senses total lightning
events, which are processed into lightning flashes by a series of filters and a clustering
algorithm.4 In order to validate the GLM data, we need to compare GLM flashes to data from
several ground-based lightning detection networks, in order to encompass GLM’s nearly hemi-
spheric FOV, as shown in Fig. 1. The ground truth systems we used in this study are described

*Address all correspondence to Monte Bateman, E-mail: monte.bateman@nasa.gov

1931-3195/2020/$28.00 © 2020 SPIE

Journal of Applied Remote Sensing 032406-1 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

Fig. 1 FOV for the GLM on GOES-East. The blue line indicates the GLM FOV. The GLM FOV
covers most of the CONUS, Central America, and South America. All data for our analysis are
from the GLM on GOES-East.

below. Lightning flashes are typically composed of several strokes (about three on average).5
A challenge in merging the different reference data sets is that some ground-based lightning
detection networks report strokes, others report flashes, and some report both.
The first system we used is Earth Network’s Global Lightning Network (ENGLN).6 This is a
source that combines the Earth Network’s Total Lightning Network (ENTLN)7 with the World
Wide Lightning Location Network (WWLLN).8,9 ENTLN provides high DE stroke data (what
they call “portions”) that cover the continental United States (CONUS) and northern and eastern
South America. The WWLLN adds global coverage, but at a much lower DE (10% to 20%).10
These coverages are apparent from the maps that will be presented.
The next set of systems are made and operated by Vaisala. The Global Lightning Dataset
GLD36011 provides wide area, near global coverage, and high DE lightning stroke data.12,13
The data purchased for this study are limited by contract to between −40 and −150 longitude.
These limits are apparent in the maps that will be presented. Next, we used the National
Lightning Detection Network (NLDN) stroke and flash data.14,15 These provide high DE
cloud-to-ground lightning data over CONUS and to about 100 km beyond the shores of
CONUS. Finally, to fill in the northern regions of North America, we use the Canadian
Lightning Detection Network.16 This network provides lightning flash data with sensors
similar to the NLDN.14

3 Methodology

3.1 Creating a “Virtual” Lightning Network


One of the difficulties in calculating DE for the GLM is finding an appropriate ground truth data
set.17,18 The GLM has a much wider FOV and comparable DE to the available ground truth data
sets. So to have the best possible reference data set, we synthesized a “virtual” lightning detec-
tion network from the available ground-based networks. All of these networks detect radio fre-
quency (RF) signals from lightning. We combined data from the aforementioned list of networks
using a clustering technique similar to that used in clustering GLM events into flashes.1,4,19 All
datasets are first combined and time sorted, then all sources that are within 330 ms and 16.5 km

Journal of Applied Remote Sensing 032406-2 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

of the first source are clustered into a flash. As a flash is being clustered, the time and space
extents are allowed to grow, so any subsequent event is added in if it is within 330 ms and
16.5 km of any other event already in the cluster. Our clustering can sometimes group lightning
events together that another system may have declared to be two or three separate flashes. Since
each system has its own definition of what a flash is, and because of the different types of sensors
(optical versus RF), we generally do not get a 1:1 correspondence with flashes from different
systems. This is not important for this study. We are simply looking at DE and FAR; our scheme
seeks common detections, not 1:1 accounting.
Using the clustering technique aforementioned, we clustered sources from our list of ground-
based networks into “virtual” flashes. These virtual flashes were then used as a single ground-
truth network and compared with GLM flashes. We then calculated “hits” and “misses” for
GLM and for the virtual network. Those numbers are used to calculate the DE and FAR values
(detailed as follows).

3.2 Calculation Details


Flashes are considered to be coincident between GLM and the virtual network if they are within
50 km and ±1 s. We allowed these time and space windows to be fairly “loose,” owing to
possible differences in timing and geolocation between systems. Again, the GLM is an optical
detector, sensing the optical output from the top of the cloud, while the systems that make up the
virtual network are detecting RF signals from various parts of the lightning channels. Therefore,
determination of the flash location and exact time could be quite different between these two
types of sensors.
The parameters we use to calculate DE and FAR are listed here and detailed in Table 1.

A: Flashes detected by GLM, not detected by the virtual network;


B: Flashes detected by GLM, coincident with the virtual network;
C: Flashes detected by the virtual network, coincident with GLM;
D: Flashes detected by the virtual network, not detected by GLM.
It may seem that B should be equal to C. There are two reasons why this is not the case. First,
the GLM and virtual network detect different aspects of the lightning occurence (optical and RF).
Hence, they see different parts of the lightning flash, and thus, there is not (generally) agreement
on what constitutes a flash between these two types of sensors. So, we do not necessarily get
a 1:1 correspondence for flashes (or strokes) when comparing GLM (optical) with the virtual
network (RF). Second, there is no “truth” here—we are comparing two imperfect systems, so
this is a “relative” DE. It means that “who is right” depends on your point of view. Given these
definitions and conditions, we define DE as

C
DE ¼ ;
ðC þ DÞ
EQ-TARGET;temp:intralink-;sec3.2;116;265

and FAR as
A
FAR ¼ :
ðA þ BÞ
EQ-TARGET;temp:intralink-;sec3.2;116;217

Table 1 A 2 × 2 table illustrating the definitions of ABCD and showing the relationship between
GLM and the virtual network coincidences and noncoincidences (“hits” and “misses”).

“Other” missed Coincidence

GLM A = Virtual net miss B = GLM coincident with virtual net

Virtual network D = GLM miss C = Virtual net coincident with GLM

Journal of Applied Remote Sensing 032406-3 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

DE is defined as the number of flashes detected that are coincident to GLM and the virtual
ground truth system, divided by the total number of flashes detected by the virtual ground truth
system. FAR is defined as the number of flashes detected by GLM that are not coincident with
the virtual ground truth system, divided by the total number of GLM flashes. We understand that
these are not absolute (but instead relative) DE and FAR, as neither of the systems is 100%
efficient.
DE and FAR are functions of time and space, so the answer to “what is the DE or FAR”
depends on time and location. So each plot requires choices to be made: integrate over X spatial
scale and T time interval. Larger distances and longer time intervals will naturally “wash out”
small scale detail; conversely, smaller distances and shorter time periods increase the noise in the
results. For the GLM specification, the DE and FAR are defined to be evaluated over a 24-h
period.
In order to evaluate the entire GLM FOV, we have chosen a grid that consists of spatial boxes
that are 1 deg × 1 deg latitude/longitude, and we integrate time over many months. For GLM on
GOES-16, the data were declared “fully validated for operational use” in January 2018, so we
analyzed data from January 1, 2018, to May 31, 2019 (17 months). Note that there were many
algorithm changes to the processing software over these months,3 which are highlighted in
Table 2. We will, for now, not attempt to determine the exact effect of each change.
Filtering: Since the GLM FOV is very large, and there is not much ground truth data over a
large fraction of that FOV, we have implemented a filtering function that prevents wild fluctua-
tions in the DE and FAR statistics owing to small flash counts in some grid boxes. This filter

Table 2 Details on the cluster-filter algorithm fixes and upgrades during the period of this study.
A “hot” pixel is one that consistently outputs data, regardless of input.

Change # Date Description Affects DE/FAR?

PR.06.02.05 January 10, 2018 Fixed the second-level threshold filter No

DO.06.03.00 June 19, 2018 Removed radiation “dots,” fixed scale factor in Yes
netCDF files

DO.07.00.00 October 15, 2018 (1) Use adjusted event times in L2 product No

(2) Group and flash areas fixed

(3) Look-up table values updated

(4) Orphan and childless events and groups


fixed

PR.07.01.00 October 29, 2018 (1) Update GLM lookup table Yes

(2) Updated lightning ellipsoid values

(3) New second-level threshold filter to mitigate


“Bahama Bar”

PR.07.02.00 November 5, 2018 Handling “burst events” that were causing high Yes
numbers of false events in the L1b file, and
causing the L2 files to be empty

PR.07.03.00 November 15, 2018 Updated second level threshold table to handle Yes
“hot” pixels

PR.07.08.00 February 27, 2019 (1) Second-level filter code change Yes

(2) Data burst filter code change

PR.07.10.05 April 30, 2019 (1) Threshold changes to mitigate noise Yes

(2) Turn on contrast leakage filter

(3) Update second-level threshold to better


handle “Bahama Bar”

Journal of Applied Remote Sensing 032406-4 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

Fig. 2 Flash counts for the GLM and virtual ground network. Note that any grid box with fewer than
20 GLM or Virtual Network flashes are not considered and are therefore blank (white). During
these 17 months, GLM saw 517 million flashes, while the virtual ground network saw 738 million
flashes.

disregards data in which any grid box has <20 flashes from either data source (see Fig. 2). The
number 20 was chosen for this study after determining how many flashes typically accumulate
over this many months-long time period to assure that we were looking at signal, and not
just noise.

4 Results and Discussion

4.1 DE Maps
The color scale for Figs. 3 and 4 is as follows: >70% DE is green, 50% to 70% DE is dark green,
25% to 50% DE is light gray, and <25% DE is dark gray. The yellow color indicates that GLM
detected lightning but the virtual ground systems did not. If the virtual ground systems detected
lightning and GLM did not, the color for that grid box is red.
Shown in Fig. 3 is GLM DE relative to the virtual network (24 h) for the whole data analysis
period (January 1, 2018, to May 31, 2019). In this figure, we see that most of the FOV for the
GLM is green, which indicates >70% DE, the DE specification for GLM. We see the decreased
DE in northwest CONUS, which has been reported by others previously.20 The red around the
edges of the plot occurs because it is outside the FOV of the GLM, and therefore there are no
GLM detections there. The red and yellow areas west of South America and at the southern end
of the plot will be addressed in the next paragraph.
GLM DE plotted against the virtual network for day and night is shown in Fig. 4(a) and 4(b),
respectively. The area west of South America is where GLM is detecting lightning (or noise), but
the virtual ground system is not. Notice that most of this feature disappears in the night time plot.
This indicates that whatever is happening there is a daytime phenomenon and is likely glint off
the ocean’s surface that is misclassified as lightning. For the area in the southern part of the plot,
the reverse is true; the feature increases at night. This suggests that this phenomenon is caused by
lowering the GLM detection threshold at night. Note that there is very little virtual ground truth
data in that part of the GLM FOV.

4.2 FAR Maps


Figures 5 and 6 show the FAR for our dataset. The color scale for these plots is as follows: <5%
FAR (best) is light gray, 5% to 30% is dark gray, 30% to 60% is dark orange, and >60% is bright

Journal of Applied Remote Sensing 032406-5 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

Fig. 3 GLM DE: 17 months, January 1, 2018, to May 31, 2019. Mostly green, which meets the DE
specification. Yellow area west of South America, and far south, likely due to a lack of ground
truth data in those areas. This lack of ground truth data can be seen in Fig. 2.

Fig. 4 GLM DE: same time period; (a) daytime and (b) nighttime. Notice that almost all of the
yellow area west of South America disappears from day to night. Also notice that the far southern
yellow is greater at night. We speculate this is because the GLM thresholds are decreased at night,
and we have much more data. Again there is very little ground truth at the southern regions of
South America (see Fig. 2).

orange. We have also filtered out any grid square with <20 total GLM or virtual network flashes
from this calculation.
GLM FAR relative to the virtual network (24 h) is shown in Fig. 5. The first thing to notice
is that where we have the best ground truth data, in CONUS, we get the best FAR. The dark
gray regions are telling us where we have the most ground truth data: North America and

Journal of Applied Remote Sensing 032406-6 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

Fig. 5 GLM FAR: 17 months, January 1, 2018, to May 31, 2019. The regions of better ground truth
data are apparent; around CONUS and eastern South America, centered on Brazil. Notice that a
large part of CONUS is light gray, which is better than 5% FAR, meeting specification. Also notice
the bright orange “Bahama Bar” out to the east of CONUS; this artifact is discussed in the text.
Other areas in the outer reaches of the FOV are brighter orange due to lack of ground truth data
there (see Fig. 2). Also notice the vertical line just off the coast of Brazil. This is where our GLD360
data feed terminates by contract.

Fig. 6 GLM FAR: same time period: (a) daytime and (b) nighttime. The horizontal bar artifacts
previously mentioned are only a daytime phenomenon. Otherwise, these panels are quite similar.
Note that the night time FAR is better. We knew that GLM DE at night was higher, so the false
alarm ratio should be better as well, and we are seeing that. Again the high FAR values are out
over the ocean, far from land.

Journal of Applied Remote Sensing 032406-7 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

eastern and northern South America. There is also quite plainly an artifact east of North
America (referred to as the “Bahama Bar”); the source of this has been determined to be glint
and blooming related4 and should be mostly mitigated in the current update of the GLM-
processing code. A similar horizontal stripe can be seen across the central part of South
America. There is also a vertical line in the eastern part of this plot; that is where our
GLD360 data end, limited by contract. Anywhere the ground truth data becomes sparse or
nonexistent, the plot shows bright orange.
GLM FAR relative to the virtual network (day/night) is shown in Fig. 6. This plot splits the
FAR results into daytime [Fig. 6(a)] and nighttime [Fig. 6(b)]. The horizontal bars mentioned
previously apparently are only a daytime phenomenon, further supporting the glint/blooming
source. Otherwise, these two panels are fairly similar. The area of “best FAR” (light gray)
in CONUS is larger in the daytime than at night, which is surprising because the DE is better
at night. The GLM threshold is set lower at night, which might account for the higher FAR. The
area of dark gray in South America is larger at night.

5 Concluding Remarks
The total DE map for GLM (Fig. 3) is mostly in excess of the 70% requirement specification
(green). The average DE over the whole FOV, over the whole dataset is 0.77. Note that this value
includes grid boxes that are outside of the FOVof the GLM (shown in red). Even so, GLM easily
meets the specification for DE, which is 70% across the whole FOV. For the few areas that are not
green, we are continuing to investigate. The conclusion we draw from this result is that the GLM
on GOES-East (GLM-16) is producing high-quality, (high DE) data, over nearly its entire FOV
and easily meets the 70% over 24 h specification requirement.
The day/night plots (Fig. 4) show that the GLM performs well during the daytime and even
better at night. The one area of yellow and red to the west of South America disappears at night,
leading us to believe that this is a daytime phenomenon, perhaps ocean glint and/or blooming.
Although the specification for the GLM DE is for a 24-h period, GLM also meets that speci-
fication for both day (0.82) and night (0.73). The individual areas where the DE falls below the
specification warrant further study.
The FAR maps (Figs. 5 and 6) show that where we have the best quality ground truth data, the
GLM performs well. In areas of the Earth where we have minimal (or no) ground truth data, the
FAR is high, as expected. Our results indicate that the FAR specification of 5% over 24 h is not
met for a large fraction of the GLM FOV. Given that the areas where the FAR is well above 5%
are also the areas where we have the least amount of data from the virtual ground truth network,
much of the “failure” of GLM meeting the specification may be due to simply the lack of cor-
roborating “ground truth” data.
In Fig. 5, the regions of better ground truth data are apparent; around CONUS and eastern
South America, centered on Brazil. Notice that a large part of CONUS is light gray, which indi-
cates better than 5% FAR. Also notice the bright orange bar out to the east of CONUS; the source
of this artifact has been determined to be glint and blooming related and should be fixed in the
current update of the GLM-processing code.21 Other areas in the outer reaches of the FOV are
brighter orange because of the lack of ground truth data. Also notice the vertical edge between
the bright orange and dark orange just off the coast of Brazil. This is where our GLD360 data
feed terminates by contract.
Figure 6 shows daytime and nighttime FAR. The horizontal bar artifacts previously men-
tioned disappear at night. Otherwise, these panels are quite similar. Note that the nighttime
FAR is better. The GLM DE should be better at night (lower thresholds and less background
noise), so the FAR should be better as well, and we are seeing that. Again the high FAR values
are out over the ocean, far from land where there is a lack of data in our virtual ground truth
network.
Our results indicate that currently the GLM is meeting the overall specification for DE,
although certain storms in certain situations have lower DE.20 However, GLM is not yet meeting
the FAR specification. Algorithm improvements are being made that eliminate much of the glint
and blooming artifacts,4,21 and further work is needed to address the FAR issue.

Journal of Applied Remote Sensing 032406-8 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

Acknowledgments
This work was supported by NASA 80MSFC17M0022, Cooperative Agreement with
Universities Space Research Association, and NASA Research Opportunities in Space and
Earth Science, Grant No. NNX17AJ10G U.S. and European Geostationary Lightning Sensor
Cross-Validation Study. All the data used in this paper are available at the Global
Hydrology Resource Center (GHRC) Distributed Active Archive Center (DAAC).22

References
1. S. J. Goodman et al., “The GOES-R geostationary lightning mapper (GLM),” Atmos. Res.
125-126, 34–49 (2013).
2. “GOES-R series data book,” CDRL PM-14 Rev A., NOAA-NASA, 2019, https://www
.goes-r.gov/downloads/resources/documents/GOES-RSeriesDataBook.pdf.
3. S. D. Rudlosky et al., “Initial geostationary lightning mapper observations,” Geophys. Res.
Lett. 46, 1097–1104 (2019).
4. D. M. Mach, “Science and operational geostationary lightning mapper cluster algorithms
comparison,” in Poster AE24A-02, AGU Fall Meeting, San Francisco, California
(2019).
5. M. A. Uman, The Lightning Discharge, Academic Press, San Diego, California (1987).
6. S. Heckman, “ENTLN status update,” in XV Int. Conf. Atmos. Electr., Norman, Oklahoma
(2014).
7. C. Liu and S. Heckman, “The application of total lightning detection and cell tracking for
severe weather prediction,” in Conf. Meteorol. and Environ. Instrum. and Methods of Obs.,
Helsinki, Finland, WMO, vol. P2(7) (2010).https://www.wmo.int/pages/prog/www/IMOP/
publications/IOM-104_TECO-2010/P2_7_Heckman_USA.pdf.
8. http://wwlln.net.
9. R. L. Dowden, J. B. Brundell, and C. J. Rodger, “VLF lightning location by time of group
arrival (TOGA) at multiple sites,” J. Atmos. Sol.-Terr. Phys. 64, 817–830 (2002).
10. R. E. Burgesseur, “Assessment of the world wide lightning location network (WWLLN)
detection efficiency by comparison to the lightning imaging sensor (LIS),” Q. J. R.
Meteorol. Soc. 143, 2809–2817 (2017).
11. S. Mallick et al., “Evaluation of the GLD360 performance characteristics using rocket and
wire triggered lightning data,” Geophys. Res. Lett. 41, 3636–3642 (2014).
12. R. K. Said, M. B. Cohen, and U. S. Inan, “Highly intense lightning over the oceans:
estimated peak currents from global GLD360 observations,” J. Geophys. Res.: Atmos.
118, 6905–6915 (2013).
13. R. K. Said, U. S. Inan, and K. L. Cummins, “Long-range lightning geolocation using a VLF
radio atmospheric waveform bank,” J. Geophys. Res. 115, D23108 (2010).
14. R. E. Orville et al., “The North American Lightning Detection Network (NALDN): analysis
of flash data: 2001-09,” Mon. Weather Rev. 139, 1305–1322 (2011).
15. K. L. Cummins et al., “A combined TOA/MDF technology upgrade of the U.S. National
Lightning Detection Network,” J. Geophys. Res. 103(D8), 9035–9044 (1998).
16. W. R. Burrows et al., “Lightning occurrence patterns over Canada and adjacent
United States from lightning detection network observations,” Atmos.-Ocean 40, 59–80
(2002).
17. J. P. Charba et al., “Lamp upgraded convection and total lightning probability and
“potential” guidance for the conterminous United States,” Weather Forecasting 34,
1519–1545 (2019).
18. C. J. Schultz et al., “Characteristics of lightning within electrified snowfall events using
lightning mapping arrays,” J. Geophys. Res. 123(4), 2347–2367 (2018).
19. D. M. Mach et al., “Performance assessment of the optical transient detector and lightning
imaging sensor,” J. Geophys. Res. 112, D09210 (2007).
20. K. Cummins, “Assessment of the GLM-16 & 17 level 2 dataset: July 20–31, 2019,” in GLM
Sci. Meeting, Huntsville, Albama (2019). https://goes-r.nsstc.nasa.gov/home/meeting-agenda-
2019.

Journal of Applied Remote Sensing 032406-9 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Bateman and Mach: Preliminary detection efficiency and false alarm rate assessment. . .

21. C. Tillier and S. Edgington, “Recent changes to GLM ground processing,” in GLM Sci.
Meeting, Huntsville, Albama (2019), https://goes-r.nsstc.nasa.gov/home/meeting-agenda-
2019.
22. https://ghrc.nsstc.nasa.gov/home/.

Monte Bateman is a senior scientist at the Universities Space Research Association at NASA’s
Marshall Spaceflight Center. He received his BS and MS degrees in engineering physics from
the University of Oklahoma in 1983 and 1987, respectively, and his PhD in physics from
the University of Oklahoma in 1992. He is a member of the GLM calibration working group
and the GLM calibration/validation team. His current research interests include lightning and
thunderstorm physics, instrumentation, and lightning and severe weather.

Douglas Mach received his BS and MS degrees in engineering physics from the University of
Oklahoma in 1982 and 1984, respectively, and his PhD in physics from the University of
Oklahoma in 1987. He is a member of the Algorithm Working Group and Calibration Working
Group for the GLM. He is a member of the EUMETSAT Lightning Mapper Mission Advisory
Group. His current research interests include electronic instrumentation, scientific software,
storm electricity, remote sensing, lightning physics, lightning hazards to launch vehicles, opera-
tional research and analysis, unmanned aerial vehicle measurements, satellite measurements of
lightning, and international scientific cooperation.

Journal of Applied Remote Sensing 032406-10 Jul–Sep 2020 • Vol. 14(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

You might also like