A Comparison of Two Ground-Based Lightning Detection Networks against the Satellite-Based Lightning Imaging Sensor (LIS)

Kelsey B. Thompson Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by Kelsey B. Thompson in
Current site
Google Scholar
PubMed
Close
,
Monte G. Bateman Universities Space Research Association, Huntsville, Alabama

Search for other papers by Monte G. Bateman in
Current site
Google Scholar
PubMed
Close
, and
Lawrence D. Carey Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by Lawrence D. Carey in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Lightning stroke data from both the World Wide Lightning Location Network (WWLLN) and the Earth Networks Total Lightning Network (ENTLN) were compared to lightning group data from the Lightning Imaging Sensor (LIS) from 1 January 2010 through 30 June 2011. The region of study, from 39°S to 39°N latitude, chosen based on the orbit of LIS, and 164°E east to 17°W longitude, chosen to approximate the possible Geostationary Lightning Mapper (GLM) longitude, was considered in its entirety and then divided into geographical subregions. Over this 18-month time period, WWLLN had an 11.0% entire region, 13.2% North American, 6.2% South American, 16.4% Atlantic Ocean, and 18.9% Pacific Ocean coincidence percent (CP) value. The ENTLN CP values were 28.5%, 63.3%, 2.2%, 3.0%, and 2.5%, respectively. During the 18 months, WWLLN CP values remained rather consistent but low and often higher over ocean than land; ENTLN CP values showed large spatial and temporal variability. With both networks, North America had less variability during summer months than winter months and higher CP values during winter months than summer months. The highest ENTLN CP values were found in the southeastern United States, especially in a semicircle that extended from central Oklahoma, through Texas, along the northern Gulf of Mexico, across southern Florida, and along the U.S. East Coast. There was no significant change in CP values over time; the lowest monthly North American ENTLN CP value was found in June 2011 at 48.1%, the last month analyzed. These findings are consistent with most ENTLN sensors being located in the United States.

Corresponding author address: Kelsey Thompson, Department of Atmospheric Science, University of Alabama in Huntsville, 320 Sparkman Dr., Huntsville, AL 35805. E-mail: kelsey.thompson@nsstc.uah.edu

Abstract

Lightning stroke data from both the World Wide Lightning Location Network (WWLLN) and the Earth Networks Total Lightning Network (ENTLN) were compared to lightning group data from the Lightning Imaging Sensor (LIS) from 1 January 2010 through 30 June 2011. The region of study, from 39°S to 39°N latitude, chosen based on the orbit of LIS, and 164°E east to 17°W longitude, chosen to approximate the possible Geostationary Lightning Mapper (GLM) longitude, was considered in its entirety and then divided into geographical subregions. Over this 18-month time period, WWLLN had an 11.0% entire region, 13.2% North American, 6.2% South American, 16.4% Atlantic Ocean, and 18.9% Pacific Ocean coincidence percent (CP) value. The ENTLN CP values were 28.5%, 63.3%, 2.2%, 3.0%, and 2.5%, respectively. During the 18 months, WWLLN CP values remained rather consistent but low and often higher over ocean than land; ENTLN CP values showed large spatial and temporal variability. With both networks, North America had less variability during summer months than winter months and higher CP values during winter months than summer months. The highest ENTLN CP values were found in the southeastern United States, especially in a semicircle that extended from central Oklahoma, through Texas, along the northern Gulf of Mexico, across southern Florida, and along the U.S. East Coast. There was no significant change in CP values over time; the lowest monthly North American ENTLN CP value was found in June 2011 at 48.1%, the last month analyzed. These findings are consistent with most ENTLN sensors being located in the United States.

Corresponding author address: Kelsey Thompson, Department of Atmospheric Science, University of Alabama in Huntsville, 320 Sparkman Dr., Huntsville, AL 35805. E-mail: kelsey.thompson@nsstc.uah.edu

1. Introduction

The Geostationary Lightning Mapper (GLM), scheduled to launch aboard the Geostationary Operational Environmental Satellite R series (GOES-R) in early 2016, is a satellite-based optical lightning detection sensor (http://www.goes-r.gov/; Goodman et al. 2013). To test data processing and product-generation algorithms for GLM, a high-fidelity set of proxy data is needed. The Lightning Imaging Sensor (LIS) is the best analog for such data, but it only observes an area for a short time because of its low-Earth orbit (LEO). Early proxy data were based on Lightning Mapping Array (LMA) data. These ground-based systems have high detection efficiencies (DE), but they only detect lightning in a limited area (e.g., Rison et al. 1999; Krehbiel et al. 2000; Koshak et al. 2004; Thomas et al. 2004; Goodman et al. 2005). Ground-based lightning detection networks with broad coverage are needed to develop and test many applications. Such networks exist, but their performance abilities are not well known. Goals of this study were to learn the performance ability of the World Wide Lightning Location Network (WWLLN) and the Earth Networks Total Lightning Network (ENTLN), allowing their data to be used appropriately in the many research and operational applications that have begun using them, and to make a detailed comparison with LIS as a first step toward generating larger-scale GLM proxy datasets.

2. Data sources

a. LIS

LIS is an instrument aboard the Tropical Rainfall Measuring Mission (TRMM) satellite, launched in November 1997 (Christian et al. 1999). The charge-coupled device (CCD) array of 128 × 128 pixels, where each pixel maps to about 5 km × 5 km, gives a viewing area of about 640 km × 640 km. LIS uses optical pulses to detect total [intracloud (IC) and cloud-to-ground (CG)] lightning between about 38°S and 38°N latitude as the satellite moves in LEO. The continuous movement means LIS views a location for about 80 s. The GLM is a satellite-based optical instrument similar to LIS but will be in geostationary Earth orbit (GEO). The use of similar technologies by LIS and GLM makes LIS the best analog for GLM proxy data.

As described by Mach et al. (2007), a clustering algorithm is used to classify lightning detected by LIS into events, groups, and flashes. An event is a single pixel that exceeds the background threshold in a single 2-ms time frame. A group is defined as adjacent events that occur in the same 2-ms time frame. A flash is defined as a cluster of groups within 330 ms and 5.5 km of each other.

Previous studies (e.g., Ushio et al. 1999; Thomas et al. 2000; Koshak et al. 2000a) have compared LIS detections to ground-based networks. Ushio et al. (1999) found that LIS detected 24 of 42 ground flashes and more than 80% of cloud flashes detected by the Lightning Detection and Ranging (LDAR) system. They also found mean location differences of 4.3 km for cloud flashes and 12.2 km for ground flashes. Thomas et al. (2000) found that LIS detected 108 of 128 discharges detected by the New Mexico Institute of Mining and Technology (New Mexico Tech) LMA, with the best spatial agreement when they shifted LIS locations north 6 km. Others studies (e.g., Koshak et al. 2000b; Boccippio et al. 2002; Buechler et al. 2014) have looked at the capabilities and performance of the instrument. Koshak et al. (2000b) found a 98% DE in each quadrant of the CCD array. Boccippio et al. (2002) predicted an 88 ± 9% daily flash DE.

b. WWLLN

WWLLN is a ground-based lightning detection network that continuously detects lightning using electromagnetic radiation in the very low-frequency (VLF) range, 3–30 kHz (http://wwlln.net/). Presently, at least five sensors (http://wwlln.net/) need to detect a stroke to determine its location using a long baseline, time of group arrival (TOGA) technique (Dowden et al. 2002; Rodger et al. 2009). A sensor can detect a lightning stroke several thousands of kilometers away. Note that there is less attenuation of the lightning signal in the VLF range than at higher frequencies, and thus it propagates farther. The present residual timing error for received signals needs to be ≤30 μs. The WWLLN data archive dates back to 2004 (http://wwlln.net/). During 2010, the global sensor count ranged from the mid-30s to the mid-40s, with an overall increasing trend. In early 2011, the count ranged from the mid- to low 40s, with an overall decreasing trend (R. Holzworth 2011, personal communication). Changes to WWLLN over time have affected its performance. These changes include the number and placement of sensors, the algorithm used, and the criteria for data to be considered “good.”

Previous studies, (e.g., Lay et al. 2004, 2007; Rodger et al. 2004, 2005, 2006; Jacobson et al. 2006; Abarca et al. 2010; Rudlosky and Shea 2013) have compared regional and global WWLLN detections to other ground-based networks and satellite-based instruments. Many studies noted that WWLLN tended to detect lightning with large return stroke peak currents. Early studies often found WWLLN DEs less than 3%. Abarca et al. (2010) found an increase in DE from 2006–09; 2.31% for 2006–07, 2.93% for 2007–08, and 6.19% for 2008–09. They noted that CG (IC) flash DE values were higher (lower) than the combined CG and IC DE values and that WWLLN had a 4.03-km northward and 4.98-km westward location bias. Rudlosky and Shea (2013) found a DE of 6.0% for 2009, 6.8% for 2010, 8.1% for 2011, and 9.2% for 2012. They also found an 11-km average and 10-km median location offset.

c. ENTLN

ENTLN, previously known as WeatherBug Total Lightning Network (WTLN), is a ground-based lightning detection network that continuously detects lightning using parts of the spectrum between 1 Hz and 12 MHz (Liu and Heckman 2010; ENTLN 2012). Since the signals produced by most cloud flashes become comparable to those produced by ground flashes only at higher frequencies, expanding the operating range to the high-frequency bands adds improved detection of cloud flashes to the system’s ability to detect ground flashes. Areas with low sensor density favor CG lightning detection (Liu and Heckman 2010). The first sensors date back to 2009. During 2010, hundreds of sensors existed, mostly in the United States. Throughout 2010 and after, the number and location of sensors increased (S. Heckman 2011, personal communication). Earth Networks claims that they have global detection capability (ENTLN 2012). Eight sensors initially were needed to locate the lightning strokes. Because of limited sensor availability in some areas during network testing, the minimum number required was reduced to five if the local noise level was small enough that the waveforms of the lightning signals were well defined (S. Heckman 2012, personal communication). A long baseline, time of arrival (TOA) technique is used on the waveforms. The peak current and location of a stroke are determined from the time and signal amplitude data. Multiple strokes within 10 km and 700 ms are termed a flash. Flashes with one or more return strokes are classified as a CG flash. Flashes without a return stroke are classified as an IC flash (Liu and Heckman 2010).

3. Methodology

a. Coincidences

Each day in the dataset, 1 January 2010 through 30 June 2011, was analyzed separately to find coincidences. The coincidence criteria we used were a LIS group and WWLLN stroke or LIS group and ENTLN stroke within time and space windows of ≤0.4 s, ≤0.15° latitude, and ≤0.15° longitude. These values were chosen based on a sensitivity test where we considered a range of choices and tabulated the coincidences (Fig. 1). Note that LIS, WWLLN, and ENTLN accuracy are all different. WWLLN and ENTLN coincidence percent (CP) values (discussed in section 3c) were calculated based on different time, latitude, and longitude interval sizes. Figure 1a (Fig.1b) shows the entire region 18-month WWLLN (ENTLN) CP values based on constant, equal latitude and longitude constraints and a varying time constraint. Figure 1c (Fig. 1d) shows the entire region 18-month WWLLN (ENTLN) CP values based on a constant time constraint and varying, equal latitude and longitude constraints. The entire region covers 39°S–39°N latitude, chosen based on the orbit of LIS, and 164°E east to 17°W longitude, chosen to approximate the possible GLM longitude (GOES-East and GOES-West). As the allowed time difference increased, CP values increased. As the allowed latitude and longitude difference increased, CP values also increased. We looked for the location where the curve began to flatten out and therefore where the increase in the CP value was no longer significant. We then graphed the slope midway between each tested constraint from Fig. 1 (not shown). Those graphs also showed that around 0.4 s, 0.15° latitude, and 0.15° longitude, the CP values no longer increased significantly. Near the chosen constraints, the change in CP value became less than 1.0 (1.6) and the slope became less than 10 (31) in the varying time (latitude and longitude) graphs. Note that these values corresponded to a denominator change of 0.1 s (0.05° latitude and longitude).

Fig. 1.
Fig. 1.

Entire region 18-month (a) WWLLN and (b) ENTLN CP values based on constant, equal latitude and longitude constraints (line color) and a varying time constraint. The vertical black line runs through the chosen time constraint. (c),(d) As in (a),(b) but based on a constant time constraint (line color) and varying, equal latitude and longitude constraints. The vertical black line runs through the chosen latitude and longitude constraint. In all plots, the colored × symbols mark the tested constraints, and the black ♦ symbol marks the chosen coincidence criteria.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

To not miss any possible coincidence, we used these very large coincidence “windows” for time and space; the duration of a lightning flash is typically about 0.4 s. Uman (1987) stated that the total time duration of a ground and cloud discharge (flash) is about the same, roughly half a second. The large windows meant that a coincidence could have occurred between any group in a LIS flash and any stroke in a WWLLN or ENTLN flash, so essentially we were comparing flashes. Note that our definition of a CP value is not the same as a flash DE or a flash CP value.

Recall that a LIS event is a single pixel that exceeds the background threshold in a single 2-ms time frame, a LIS group is formed when multiple adjacent pixels are illuminated in the same 2-ms time frame, and a LIS flash is a cluster of groups within 330 ms and 5.5 km of each other (Mach et al. 2007). WWLLN usually only detects one stroke in a flash. ENTLN produces stroke and flash data, but only stroke data were available during this study. We decided that out of LIS events, groups, and flashes, groups would most closely resemble WWLLN and ENTLN strokes. A LIS group is similar to a physical stroke, which is a continuously illuminated discharge within a flash. Many times, a LIS group will be exactly the same as a physical stroke. Differences come about when 1) part of the physical stroke becomes too dim to break the background threshold, which results in the stroke being separated into two or more LIS groups; 2) a physical stroke begins in one 2-ms time frame and ends in another 2-ms time frame, which results in the stroke being separated into two LIS groups; or 3) two or more physical strokes occur inside a cloud and overlap in time–space, which results in only one LIS group being recorded (D. M. Mach 2013, personal communication).

b. Redundancy

Redundancy existed because numerous coincidences between LIS groups and WWLLN strokes or LIS groups and ENTLN strokes did not have a one-to-one correspondence. When one LIS group had more than one coincident WWLLN or ENTLN stroke, we recorded redundancy of strokes per group. Then, based on the same coincidences, when one WWLLN or ENTLN stroke had more than one coincident LIS group, we recorded redundancy of groups per stroke. Since a single WWLLN or ENTLN stroke often coincided with more than one LIS group, the number of coincident strokes contains repeats. At times, redundancy was likely an artifact caused by the large time and space criteria that were used to determine a coincidence. Again, these large time and space windows were chosen such that we did not miss any possible coincidence in order to give the ground-based networks every possible benefit of the doubt.

c. CP values

Daily, monthly, seasonal, and 18-month WWLLN and ENTLN CP values with respect to LIS groups were calculated for 1) the entire region, 2) North America, 3) South America, 4) the Atlantic Ocean, and 5) the Pacific Ocean subregions. Note that the subregion boundaries are shown in all geographical figures (e.g., Fig. 2). CP values, for a chosen time period, were computed by dividing the number of LIS groups with at least one coincident WWLLN or ENTLN stroke by the total number of LIS groups. Note that a monthly or 18-month CP value is not a mean of the daily CP values, but the percent of LIS groups for each specific time period that had a coincident stroke. We chose the term CP instead of DE since DE has its own definition. Again, note that a CP value is not the same as a flash DE or a flash CP value. A CP value of 100% meant that every LIS group had at least one coincident WWLLN or ENTLN stroke and a CP value of 0% meant that no LIS group had a coincident WWLLN or ENTLN stroke. Note that redundancy did not influence CP values; a LIS group that had more than one coincident WWLLN or ENTLN stroke was not counted more than once when determining a CP value. We also calculated the means and standard deviations of daily CP values that are presented with ranges of daily CP values and time series plots.

Fig. 2.
Fig. 2.

Number of LIS groups in each 1° × 1° grid square for the 18 months. Note that the thick black lines mark the North American, South American, Atlantic Ocean, and Pacific Ocean subregion boundaries.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

d. ENTLN ground versus cloud stroke coincidences

Since ENTLN makes a judgment as to each stroke’s type, we classified each coincidence between a LIS group and an ENTLN stroke as a ground or cloud coincidence. One LIS group often had multiple coincident ENTLN strokes, or redundancy of strokes per group. Sometimes, the redundant strokes were both ground and cloud strokes. To count the number of ground and cloud coincidences, a LIS group that had numerous coincident ENTLN strokes was counted more than once. For example, a LIS group with two coincident ENTLN ground strokes and three coincident ENTLN cloud strokes was counted 5 times. Therefore, the number of LIS groups and the number of coincident ENTLN strokes contained repeated values. To calculate ground and cloud coincidence percent (GCP and CCP, respectively) values, we divided, separately, the number of ground coincidences and cloud coincidences by the total number of coincidences. Although ground and cloud coincidence counts and GCP and CCP values were influenced by redundancy, since some LIS groups had both coincident ENTLN ground and cloud strokes, recall that CP values (WWLLN and ENTLN) were not influenced by redundancy. We did not classify coincidences between LIS groups and WWLLN strokes as ground or cloud coincidences because WWLLN does not distinguish stroke type; use of the VLF band with widely separated sensors means most WWLLN detections are ground detections.

e. Large radiance LIS groups

The largest radiance LIS groups signified what LIS saw as the brightest lightning and presumably the highest currents (e.g., Idone and Orville 1985). Note that high current increases the air temperature, which ionizes more atoms and provides the light associated with lightning. We chose the 15 largest radiance LIS groups each day for North America, South America, the Atlantic Ocean, and the Pacific Ocean, and then determined if those LIS groups had a coincident WWLLN or ENTLN stroke. Fifteen LIS groups each day provided a larger sample than using just the largest radiance value each day. We did not include days with fewer than 15 total LIS groups. This showed how well the radio frequency (RF) sensors detected the “strongest” optically detected lightning. We hypothesized that the largest radiance LIS groups may radiate more RF energy and therefore be more likely detected by WWLLN and/or ENTLN.

f. Grid square figures

In all geographical figures (Figs. 2, 3, 5, and 6), the thick black lines mark the subregion boundaries. We generated both unsmoothed (not shown) and smoothed (e.g., Figs. 3 and 5) CP value figures. For each 1° × 1° grid square, we divided the number of LIS groups with a coincident WWLLN or ENTLN stroke by the total number of LIS groups to find the unsmoothed CP value. We then reduced the amount of noise with 3 × 3 smoothing. (Note that we also generated unsmoothed 2° × 2° grid square CP value figures, but those were also noisy.) The outermost latitude and longitude grid squares were not included in the smoothing process. For interior grid squares, CP values were smoothed by replacing each value in the center of a 3 × 3 grid square block with the average of the nine CP values. If one or more surrounding grid squares did not have any LIS groups, then the average was based on the number of grid squares that did have LIS groups. If no surrounding grid square had any LIS groups, then the unsmoothed value was kept. Grid squares that did not have any LIS groups were skipped. Unless otherwise noted, grid square CP values in the remaining text refer to the smoothed CP value. We also calculated the means and standard deviations of the smoothed grid square CP values. Note that a grid square that crossed subregion boundaries (since subregion boundaries were chosen to fit geographic features and were not whole-degree latitude and longitude values) was placed in the subregion corresponding to its northwestern corner latitude and longitude value. Also note that means and standard deviations of smoothed grid square CP values that are presented with the 18-month smoothed CP value figures are different than means and standard deviations of daily CP values that are presented with ranges of daily CP values and time series plots. With the grid square values, calculations were based on the smoothed CP value of each grid square in the entire region or a subregion. To generate GCP (CCP) value figures (e.g., Figs. 6c and 6d), for each 1° × 1° grid square, we divided the number of ground (cloud) coincidences by the total number of coincidences. Figures 2, 6a, and 6b show the number of LIS groups in each grid square with a certain condition. The color of a grid square indicates its number or CP value, depending on the plot type. White grid squares did not have any LIS groups.

Fig. 3.
Fig. 3.

The 3 × 3 smoothed 18-month WWLLN CP value of each 1° × 1° grid square.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

4. Findings

a. LIS group distribution

During the 18 months, LIS recorded 8.4 × 106 total groups in the entire region with most of them concentrated in North and South American land locations. Many of those grid squares had at least 1.8 × 103 LIS groups (Fig. 2).

b. Coincidences between a LIS group and a WWLLN stroke

We found the largest number of coincidences between a LIS group and a WWLLN stroke in the northwestern part of the South American subregion, southern Mexico, and the southeastern Unites States and surrounding ocean locations (not shown). Many of those grid squares had 0.45–3.6 × 103 LIS groups with at least one coincident WWLLN stroke.

In the entire region, 9.2 × 105 LIS groups had at least one coincident WWLLN stroke. This left 7.5 × 106 LIS groups without a coincident WWLLN stroke and gave the entire region an 18-month WWLLN CP value of 11.0% (Table 1). (Note that monthly CP values are discussed later and are found in Table 1 and Fig. 7.) The 9.2 × 105 LIS groups with a coincidence were associated with 1.3 × 106 WWLLN strokes, which signified redundancy. We found an 18-month WWLLN CP value of 13.2% in North America, 6.2% in South America, 16.4% in the Atlantic Ocean, and 18.9% in the Pacific Ocean. Each subregion had redundancy.

Table 1.

WWLLN and ENTLN CP values. Note that a CP value is the percent of LIS groups that had a coincident WWLLN or ENTLN stroke for each specific time period; it is not a mean of the daily CP values. CP values have a unit of percent.

Table 1.

In the 18-month WWLLN CP value figure, we found the most consistent, but low, grid square CP values in North and South American land locations (Fig. 3). Most CP values ranged from >0% to 20%, with the higher CP values located in the northwestern part of South America, southern Mexico, and the south-central United States. In North America, the mean and standard deviation of grid square CP values were 20.0 ± 11.9. In South America, those values were 8.8 ± 7.4. Compared to land grid squares, ocean grid squares had more variable and higher CP values. Many Atlantic and Pacific Ocean grid square CP values ranged from >0% to 50%, with the highest clustered CP values located near the east–west North American–Pacific Ocean subregion boundary (note that this is in the Northern Hemisphere ITCZ) and in the northern Pacific Ocean subregion. With both the Atlantic and Pacific Ocean subregions, we found higher CP values in the northern part of the subregion than in the southern part of the subregion. The mean and standard deviation of grid square CP values were 16.9 ± 10.3 in the Atlantic Ocean, 22.7 ± 14.6 in the Pacific Ocean, and 18.3 ± 13.4 in the entire region.

The left and right panels of Fig. 4 (top) show time series plots that reveal day-to-day variability of the entire region and North American WWLLN CP values, respectively. The bottom panels show the corresponding number of LIS groups each day. The entire region shows a range of 2.1%–42.5% for daily CP values and North America shows a range of 0.0% (14 days) to 100.0% (4 days) for daily CP values. The corresponding 18-month mean and standard deviation of daily CP values were 12.6 ± 5.8 in the entire region and 24.3 ± 18.1 in North America. Note the lower North American day-to-day CP value variability around day 200 and then after about day 500—approximately Northern Hemisphere summer month days. Near day 215, when almost 5.5 × 104 LIS groups were recorded in North America, the CP values were around 5%. The days with a CP value of 0.0% or 100.0%, near day 350, each had less than about 1.0 × 104 LIS groups.

Fig. 4.
Fig. 4.

(bottom) The number of LIS groups in (left) the entire region and (right) North America, and time series plots for (top) WWLLN and (middle) ENTLN show the CP value each day from 1 Jan 2010 through 30 Jun 2011. Note that because LIS did not record any groups in North America on 4 Jan 2010, 19 Jan 2011, 9 Feb 2011, 11 Feb 2011, 22 Feb 2011, or 23 Feb 2011, those days do not have a CP value and resulted in a break in the line in the time series plots.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

The time series plots and the number of LIS groups for South America, the Atlantic Ocean, and the Pacific Ocean are not shown. The range of daily CP values and 18-month mean and standard deviation of daily CP values were 0.0%–46.4% and 7.8 ± 5.1 in South America, 0.0%–93.3% and 16.2 ± 16.7 in the Atlantic Ocean, and 0.0%–78.6% and 20.4 ± 13.1 in the Pacific Ocean, respectively.

c. Coincidences between a LIS group and an ENTLN stroke

We found the largest number of coincidences between a LIS group and an ENTLN stroke in the North American subregion, especially the eastern two-thirds of the United States and adjacent ocean locations (not shown). Many of those grid squares had 0.18–1.8 × 104 LIS groups with at least one coincident ENTLN stroke.

In the entire region, 2.4 × 106 LIS groups had at least one coincident ENTLN stroke. This left 6.0 × 106 LIS groups without a coincident ENTLN stroke and gave the entire region an 18-month ENTLN CP value of 28.5% (Table 1). (Again, monthly CP values are discussed later and are found in Table 1 and Fig. 7.) Coincidences between LIS groups and ENTLN strokes in the entire region resulted in redundancy; the 2.4 × 106 LIS groups that had at least one coincident ENTLN stroke were associated with 8.4 × 106 ENTLN strokes. We found an 18-month ENTLN CP value of 63.3% in North America, 2.2% in South America, 3.0% in the Atlantic Ocean, and 2.5% in the Pacific Ocean. In each subregion, we found redundancy.

In the 18-month ENTLN CP value figure, we found the highest grid square CP values in the North American subregion, especially in a semicircle that extended from central Oklahoma, through Texas, along the northern Gulf of Mexico, across southern Florida, and along the U.S. East Coast (Fig. 5). There, we found grid square CP values of mostly 80%–90%. Inside the semicircle, we found grid square CP values of mostly 60%–80%, with many of the lowest values clustered on the eastern side. Moving outward from the semicircle, CP values tended to decrease. In Mexico, most grid squares had a CP value of 20%–60%. The North American mean and standard deviation of grid square CP values were 48.4 ± 23.7. In South America, we found many grid squares with a CP value of 0%. We found CP values of >0%–10% in the northwestern and eastern parts of the subregion and a few clusters of grid squares with CP values of 10%–20% located in the southeastern and extreme northwestern parts of the subregion. The South American mean and standard deviation of grid square CP values were 1.6 ± 3.4. In the Atlantic and Pacific Ocean subregions, we found mostly grid squares with a CP value of 0%. A few clusters of grid squares with nonzero CP values were found in the Atlantic and Pacific Ocean subregions close to the boundaries with North America, in the far southwestern part of the Pacific Ocean subregion, and near Hawaii. The mean and standard deviation of grid square CP values were 1.4 ± 3.7 in the Atlantic Ocean, 3.7 ± 11.8 in the Pacific Ocean, and 10.9 ± 21.8 in the entire region.

Fig. 5.
Fig. 5.

The 3 × 3 smoothed 18-month ENTLN CP value of each 1° × 1° grid square.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

The left and right panels of Fig. 4 (middle) show time series plots that reveal day-to-day variability of the entire region and North American ENTLN CP values, respectively. The bottom panels show the corresponding number of LIS groups each day. The entire region shows a range of 0.0% (2 days) to 81.5% for daily CP values. The corresponding 18-month mean and standard deviation of daily CP values were 24.8 ± 20.5. From about day 150 through day 250, or roughly June–August 2010, CP values remained between about 15% and 75%. North America shows a range of 0.0% (19 days) to 100.0% (12 days) for daily CP values. The 18-month mean and standard deviation of daily CP values were 61.7 ± 25.1. From about day 310 through day 420, LIS recorded less than about 1.0 × 104 groups each day. During that time, we found days with a 0.0% CP value and days with a 100.0% CP value. The figure also shows that around Northern Hemisphere summer months, the daily CP values did not vary quite so much; no 0.0% or 100.0% CP values were recorded.

The time series plots and the number of LIS groups for South America, the Atlantic Ocean, and the Pacific Ocean are not shown. The range of daily CP values and 18-month mean and standard deviation of daily CP values were 0.0%–30.5% and 3.1 ± 4.6 in South America, 0.0%–60.9% and 2.5 ± 7.7 in the Atlantic Ocean, and 0.0%–55.8% and 2.9 ± 6.5 in the Pacific Ocean, respectively.

d. ENTLN ground versus cloud stroke coincidences

When we separated coincidences between LIS groups and ENTLN strokes into ground and cloud coincidences, we found the largest number of both ground and cloud coincidences in the southeastern United States and adjacent ocean locations (Figs. 6a and 6b). There, many grid squares had at least 3.6 × 103 ground coincidences and at least 1.8 × 104 cloud coincidences. Most grid squares had a GCP value of 10%–20% and a CCP value of 80%–90% (Figs. 6c and 6d). Mexico and North American ocean locations not adjacent to land were more likely to have a ground coincidence than a cloud coincidence. Over the 18 months, North America and the entire region each had more than twice as many cloud than ground coincidences (Table 2). We found more ground than cloud coincidences in South America, the Atlantic Ocean, and the Pacific Ocean. With the corresponding GCP and CCP values, the CCP value exceeded the GCP value in the entire region and North America and the GCP value exceeded the CCP value in South America, the Atlantic Ocean, and the Pacific Ocean. Although most days followed the 18-month trends, we did find some days that exhibited opposite of that mentioned above. We also found days when all coincidences were one type.

Fig. 6.
Fig. 6.

Number of LIS groups in each 1° × 1° grid square with at least one coincident ENTLN (a) ground stroke or (b) cloud stroke, and (c) GCP and (d) CCP values of each 1° × 1° grid square for the 18 months. Note that redundancy is included in these figures.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

Table 2.

Number of ENTLN ground and cloud strokes involved in a coincidence with at least one LIS group (ENTLN GCP and CCP values with a unit of percent) for the 18 months. Redundancy is included.

Table 2.

e. Large radiance LIS groups

We found a mean of 49.6%, 34.3%, 27.6%, and 36.8% of 15 large radiance LIS groups per day had a coincident WWLLN stroke in North America, South America, the Atlantic Ocean, and the Pacific Ocean, respectively, over the 18 months. We found corresponding values of 69.4%, 8.3%, 4.5%, and 4.2% for large radiance LIS groups and ENTLN strokes.

5. Discussion

a. Coincidences between a LIS group and a WWLLN or ENTLN stroke

Over the 18 months, we found less CP value variability with WWLLN than with ENTLN. The 18-month WWLLN CP value figure (Fig. 3) showed that land grid squares, with CP values of mostly >0%–20%, varied much less and had lower values than ocean grid squares, with CP values of mostly >0%–50%; the highest standard deviation of subregion grid square CP values was found in the Pacific Ocean. The smoothing process decreased the amount of noise, especially in the ocean, where we often found an unsmoothed grid square CP value of 0% adjacent to one of 90%–100%. (Note that the overall trends in the unsmoothed and smoothed grid square CP value figures were the same; overall trends were also the same in both ENTLN figures.) From Fig. 2, many more LIS groups were located over land than over the ocean. This led to many more ocean, compared to land, smoothed grid square CP values that were based on fewer than nine unsmoothed CP values. The 18-month ENTLN CP value figure (Fig. 5) showed that the highest CP values, mostly 80%–90%, were found in the North American subregion, especially in a semicircle that extended from central Oklahoma, through Texas, along the northern Gulf of Mexico, across southern Florida, and along the U.S. East Coast. Outside the semicircle, as the distance from land increased, CP values steadily decreased. The CP values of mostly 20%–60% in Mexico were lower than CP values of mostly 50%–90% in the United States. The highest standard deviation of subregion grid square CP values was found in North America. We found numerous grid squares with a CP value of 0% in South America, the Atlantic Ocean, and the Pacific Ocean. This is consistent with most ENTLN sensors being located in the United States.

South America had the lowest 18-month CP value for both WWLLN and ENTLN (Table 1). We found coincidences between LIS groups and ENTLN strokes in the northwestern part of the subregion during each of the 18 months, but only found coincidences in the eastern part of the subregion from December 2010 onward. We assume that ENTLN sensors were added to the eastern part of South America in December 2010 and early 2011. Monthly South American ENTLN CP values, however, did not significantly change with time after that.

Because of less attenuation, a lightning signal propagates farther at lower frequencies than at higher frequencies and low frequencies propagate farther over the ocean than over land. This likely contributed to larger WWLLN CP values in the Atlantic and Pacific Oceans than in North and South America. Recall that WWLLN scientists state that their sensors, which detect in the VLF range, can detect a lightning stroke several thousands of kilometers away (e.g., Dowden et al. 2002; Rodger et al. 2009).

Recall from Fig. 2 that much more lightning occurred over land than over the ocean. This means that compared to ocean coincidences, land coincidences had a larger impact on WWLLN and ENTLN entire region (and North and South American subregion) CP values.

As seen in the ENTLN time series plot for the entire region, Fig. 4, and in Fig. 7 and Table 1, we found higher entire region ENTLN CP values associated with Northern Hemisphere summer months (June–August 2010 and June 2011, when many LIS groups were located in North America) than with Southern Hemisphere summer months (January–February 2010 and December 2010–February 2011, when many LIS groups were located in South America). This shows that entire region CP values were largely impacted by coincidences in North America, and it follows that we found many more coincidences between LIS groups and ENTLN strokes in North America than in South America.

Fig. 7.
Fig. 7.

Bar graphs that show monthly (a) WWLLN and (b) ENTLN CP values.

Citation: Journal of Atmospheric and Oceanic Technology 31, 10; 10.1175/JTECH-D-13-00186.1

There appeared to be a seasonal impact on North American WWLLN and ENTLN CP values. With both networks, we found less daily CP value variability and lower monthly CP values during summer months (June–August 2010 and June 2011) than winter months (January–February 2010 and December 2010–February 2011) and more daily CP value variability and higher monthly CP values during winter months than summer months (Table 1; Figs. 4 and 7). Summer months had maximum CP values of 9.7% for WWLLN and 64.0% for ENTLN, while winter months had minimum CP values of 21.7% and 64.5%, respectively. The highest 18-month standard deviation of daily CP values was also found in North America for WWLLN and ENTLN. Winter months often corresponded to relatively few total LIS groups. Of the summer months, July 2010 had the fewest total LIS groups and the highest WWLLN and ENTLN CP values. We found the lowest monthly North American WWLLN CP value in June 2010 at 5.5% and the lowest monthly North American ENTLN CP value in June 2011 at 48.1%.

b. ENTLN ground versus cloud stroke coincidences

When we counted ground and cloud coincidences based on coincidences between LIS groups and ENTLN strokes, we often found more ground than cloud coincidences in South America, the Atlantic Ocean, and the Pacific Ocean, and more cloud than ground coincidences in North America and the entire region (Fig. 6). Earth Networks states that it detects more cloud strokes when sensor density is high (Liu and Heckman 2010). From our analysis, this follows Earth Networks’ statement that most of its sensors are located in the United States (S. Heckman 2011, personal communication). Across the entire region, we found the highest number of both ground and cloud coincidences in the southeastern United States, although the number of cloud coincidences (and CCP value) exceeded the number of ground coincidences (and GCP value) by a factor of 5. Recall that redundancy is included in ground and cloud coincidences. In the eastern part of the South American subregion, the location where sensors were likely added in December 2010 and early 2011, we found that the location of ground coincidences extended to the north and south of the location of cloud coincidences. We found very few cloud coincidences in the Atlantic and Pacific Oceans and found more ground than cloud coincidences as the distance from North American land increased. This result agrees with needing a high density of ENTLN sensors for cloud stoke detection. Note that as the distance from sensors increases, the amplitude of VLF signals decreases much less rapidly than that of high frequency (HF) signals. The amplitude of signals radiated by ground flashes tends to be much larger than that of signals radiated by cloud flashes, and the radiated spectrum from ground flashes tends to have a much larger component at VLF than at HF, while signals from cloud flashes do not (e.g., Taylor 1972).

c. Large radiance LIS groups

Large radiance LIS groups did not necessarily have a coincident stroke. Those in North America had a coincident WWLLN or ENTLN stroke more often than those in the other subregions had a coincident stroke. North America was also the subregion where we found the most coincidences between any LIS group and an ENTLN stroke, but with WWLLN, we often found higher CP values in the Atlantic and Pacific Oceans than in North America. In each subregion, the 18-month (but not every monthly or daily) WWLLN and ENTLN mean percent of large radiance LIS groups with a coincident stroke exceeded the corresponding CP value based on all LIS groups. For the 18 months, that increase ranged from 11.2 to 36.4 percentage points for WWLLN (which tends to detect lightning with large return stroke peak currents) and from 1.5 to 6.1 percentage points for ENTLN. With ENTLN, in North America there was a 3.6–24.3 percentage point range of increase for each of the first 10 months, but a 1.2–20.4 percentage point range of decrease for each of the last 8 months. We hypothesized that large radiance LIS groups might be associated with more current and radio frequency energy and therefore result in an increased number of coincidences, but we still found that many large radiance LIS groups in each subregion did not have a coincident stroke. This implies that large radiance LIS groups do not necessarily produce large amplitude signals in the operating frequency bands of the ground-based networks.

6. Concluding remarks

Over the 18 months, we found temporal and spatial variability in CP values with both WWLLN and ENTLN. WWLLN CP values remained rather consistent, but low. We often found higher WWLLN CP values over the ocean than over land, but the 18-month CP values in both the Atlantic and Pacific Oceans remained below 20%, at 16.4% and 18.9%, respectively. ENTLN had high temporal variability and no significant change over the 18 months even though Earth Networks states that more sensors were added; we found the lowest monthly North American CP value in June 2011 at 48.1%, the last month analyzed. ENTLN often had much higher CP values in North America than in any other subregion; 63.3% in North America for the 18 months compared to no higher than 3.0% for the other subregions. Our results are consistent with most sensors being located in the United States (and Earth Networks has stated this), but even there we found spatially varying CP values. We found the highest ENTLN CP values in a semicircle that extended from central Oklahoma, through Texas, along the northern Gulf of Mexico, across southern Florida, and along the U.S. East Coast. Outside the semicircle, ENTLN CP values steadily decreased. Over the 18 months, we found a wide range of day-to-day WWLLN and ENTLN CP values in the entire region and each subregion. In North America, with both WWLLN and ENTLN, we found less daily CP value variability and lower monthly CP values during summer months than winter months and more daily CP value variability and higher monthly CP values during winter months than summer months. A grid square with a CP value of 0% adjacent to a grid square with a CP value of 90%–100%, seen especially in ocean locations in the unsmoothed CP value figures, corresponded to grid squares with relatively few LIS groups. With coincidences between LIS groups and ENTLN strokes, North America (again, the location with the most ENTLN sensors) was the only subregion where we found a larger 18-month cloud than ground coincidence count. Most of the cloud coincidences were concentrated in the southeastern United States, which was also the location with the largest number of ground coincidences. As the distance from land increased, ground coincidences outnumbered cloud coincidences. Clusters of cloud coincidences were also found in southeastern Brazil (where sensors were likely added beginning in December 2010) and near Hawaii. Large radiance LIS groups, which we thought might be associated with more current and radio frequency energy and therefore result in an increased number of coincidences with WWLLN and/or ENTLN strokes, only had a coincident stroke about half of the time. Thus, they do not necessarily produce larger amplitude signals in the operating frequency bands of WWLLN and ENTLN.

Acknowledgments

This work was funded by the National Oceanic and Atmospheric Administration (NOAA) Geostationary Operational Environmental Satellite R series (GOES-R) Algorithm Working Group (AWG). We thank NOAA AWG Manager Jaime Daniels and senior (chief) scientist for the GOES-R System Program, Steven Goodman, for their guidance throughout all phases of this work effort. The authors wish to thank the World Wide Lightning Location Network (http://wwlln.net), a collaboration among over 50 universities and institutions, for providing the lightning location data used in this paper. We thank Earth Networks for its collaboration and for providing ENTLN data.

REFERENCES

  • Abarca, S. F., Corbosiero K. L. , and Galarneau T. J. Jr., 2010: An evaluation of the Worldwide Lightning Location Network (WWLLN) using the National Lightning Detection Network (NLDN) as ground truth. J. Geophys. Res., 115, D18206, doi:10.1029/2009JD013411.

    • Search Google Scholar
    • Export Citation
  • Boccippio, D. J., Koshak W. J. , and Blakeslee R. J. , 2002: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. Part I: Predicted diurnal variability. J. Atmos. Oceanic Technol., 19, 13181332, doi:10.1175/1520-0426(2002)019<1318:PAOTOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Buechler, D. E., Koshak W. J. , Chrisitan H. J. , and Goodman S. J. , 2014: Assessing the performance of the Lightning Imaging Sensor (LIS) using deep convective clouds. Atmos. Res.,135–136, 397403, doi:10.1016/j.atmosres.2012.09.008.

  • Christian, H. J., and Coauthors, 1999: The Lightning Imaging Sensor. 11th International Conference on Atmospheric Electricity, H. J. Christian, Ed., NASA Conf. Publ. NASA/CP—1999-209261, 746749. [Available online at http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19990108601.pdf.]

  • Dowden, R. L., Brundell J. B. , and Rodger C. J. , 2002: VLF lightning location by time of group arrival (TOGA) at multiple sites. J. Atmos. Sol.-Terr. Phys., 64, 817830, doi:10.1016/S1364-6826(02)00085-8.

    • Search Google Scholar
    • Export Citation
  • ENTLN, 2012: Total Lightning Network. [Available online at http://www.earthnetworks.com/Products/TotalLightningNetwork.aspx.]

  • Goodman, S. J., and Coauthors, 2005: The North Alabama Lightning Mapping Array: Recent severe storm observations and future prospects. Atmos. Res., 76, 423437, doi:10.1016/j.atmosres.2004.11.035.

    • Search Google Scholar
    • Export Citation
  • Goodman, S. J., and Coauthors, 2013: The GOES-R Geostationary Lightning Mapper (GLM). Atmos. Res., 125–126, 3449, doi:10.1016/j.atmosres.2013.01.006.

    • Search Google Scholar
    • Export Citation
  • Idone, V. P., and Orville R. E. , 1985: Correlated peak relative light intensity and peak current in triggered lightning subsequent return strokes. J. Geophys. Res., 90, 61596164, doi:10.1029/JD090iD04p06159.

    • Search Google Scholar
    • Export Citation
  • Jacobson, A. R., Holzworth R. , Harlin J. , Dowden R. , and Lay E. , 2006: Performance assessment of the World Wide Lightning Location Network (WWLLN), using the Los Alamos Sferic Array (LASA) as ground truth. J. Atmos. Oceanic Technol., 23, 10821092, doi:10.1175/JTECH1902.1.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., Krider P. , and Boccippio D. J. , 2000a: LIS validation at the KSC-CCAFS. Eos, Trans. Amer. Geophys. Union, 81 (Fall Meeting Suppl.), Abstract A52C-05, 2000.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., Stewart M. F. , Christian H. J. , Bergstrom J. W. , Hall J. M. , and Solakiewicz R. J. , 2000b: Laboratory calibration of the Optical Transient Detector and the Lightning Imaging Sensor. J. Atmos. Oceanic Technol., 17, 905915, doi:10.1175/1520-0426(2000)017<0905:LCOTOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., and Coauthors, 2004: North Alabama Lightning Mapping Array (LMA): VHF source retrieval algorithm and error analyses. J. Atmos. Oceanic Technol., 21, 543558, doi:10.1175/1520-0426(2004)021<0543:NALMAL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Krehbiel, P. R., Thomas R. J. , Rison W. , Hamlin T. , Harlin J. , and Davis M. , 2000: GPS-based mapping system reveals lightning inside storms. Eos, Trans. Amer. Geophys. Union, 81, 2125, doi:10.1029/00EO00014.

    • Search Google Scholar
    • Export Citation
  • Lay, E. H., Holzworth R. H. , Rodger C. J. , Thomas J. N. , Pinto O. Jr., and Dowden R. L. , 2004: WWLL global lightning detection system: Regional validation study in Brazil. Geophys. Res. Lett., 31, L03102, doi:10.1029/2003GL018882.

    • Search Google Scholar
    • Export Citation
  • Lay, E. H., Jacobson A. R. , Holzworth R. H. , Rodger C. J. , and Dowden R. L. , 2007: Local time variation in land/ocean lightning flash density as measured by the World Wide Lightning Location Network. J. Geophys. Res., 112, D13111, doi:10.1029/2006JD007944.

    • Search Google Scholar
    • Export Citation
  • Liu, C., and Heckman S. , 2010: The application of total lightning detection and cell tracking for severe weather prediction. TECO-2010-WMO Tech. Conf. on Meteorological and Environmental Instruments and Methods of Observation, Helsinki, Finland, WMO, P2(7). [Available online at https://www.wmo.int/pages/prog/www/IMOP/publications/IOM-104_TECO-2010/P2_7_Heckman_USA.pdf.]

  • Mach, D. M., Christian H. J. , Blakeslee R. J. , Boccippio D. J. , Goodman S. J. , and Boeck W. L. , 2007: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. J. Geophys. Res., 112, D09210, doi:10.1029/2006JD007787.

    • Search Google Scholar
    • Export Citation
  • Rison, W., Thomas R. J. , Krehbiel P. R. , Hamlin T. , and Harlin J. , 1999: A GPS-based three-dimensional lightning mapping system: Initial observations in central New Mexico. Geophys. Res. Lett., 26, 35733576, doi:10.1029/1999GL010856.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , Dowden R. L. , and Thomson N. R. , 2004: Location accuracy of long distance VLF lightning location network. Ann. Geophys., 22, 747758, doi:10.5194/angeo-22-747-2004.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , and Dowden R. L. , 2005: Location accuracy of VLF World-Wide Lightning Location (WWLL) network: Post-algorithm upgrade. Ann. Geophys., 23, 277290, doi:10.5194/angeo-23-277-2005.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Werner S. , Brundell J. B. , Lay E. H. , Thomson N. R. , Holzworth R. H. , and Dowden R. L. , 2006: Detection efficiency of the VLF World-Wide Lightning Location Network (WWLLN): Initial case study. Ann. Geophys., 24, 31973214, doi:10.5194/angeo-24-3197-2006.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , Holzworth R. H. , and Lay E. H. , 2009: Growing detection efficiency of the World Wide Lightning Location Network. AIP Conf. Proc.,1118, 15, doi:10.1063/1.3137706.

  • Rudlosky, S. D., and Shea D. T. , 2013: Evaluating WWLLN performance relative to TRMM/LIS. Geophys. Res. Lett., 40, 23442348, doi:10.1002/grl.50428.

    • Search Google Scholar
    • Export Citation
  • Taylor, W. L., 1972: Atmospherics and severe storms. Remote Sensing of the Troposphere, V. E. Derr, Ed., NOAA, 17-1–17-17.

  • Thomas, R. J., Krehbiel P. R. , Rison W. , Hamlin T. , Boccippio D. J. , Goodman S. J. , and Christian H. J. , 2000: Comparison of ground-based 3-dimensional lightning mapping observations with satellite-based LIS observations in Oklahoma. Geophys. Res. Lett., 27, 17031706, doi:10.1029/1999GL010845.

    • Search Google Scholar
    • Export Citation
  • Thomas, R. J., Krehbiel P. R. , Rison W. , Hunyady S. J. , Winn W. P. , Hamlin T. , and Harlin J. , 2004: Accuracy of the Lightning Mapping Array. J. Geophys. Res., 109, D14207, doi:10.1029/2004JD004549.

    • Search Google Scholar
    • Export Citation
  • Uman, M. A., 1987: The Lightning Discharge.Academic Press, 377 pp.

  • Ushio, T., Driscoll K. , Heckman S. , Boccippio D. , Koshak W. , and Christian H. , 1999: Initial comparison of the Lightning Imaging Sensor (LIS) with Lightning Detection and Ranging (LDAR). 11th International Conference on Atmospheric Electricity, H. J. Christian, Ed., NASA Conf. Publ. NASA/CP—1999-209261, 738–741. [Available online at http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19990108601.pdf.]

Save
  • Abarca, S. F., Corbosiero K. L. , and Galarneau T. J. Jr., 2010: An evaluation of the Worldwide Lightning Location Network (WWLLN) using the National Lightning Detection Network (NLDN) as ground truth. J. Geophys. Res., 115, D18206, doi:10.1029/2009JD013411.

    • Search Google Scholar
    • Export Citation
  • Boccippio, D. J., Koshak W. J. , and Blakeslee R. J. , 2002: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. Part I: Predicted diurnal variability. J. Atmos. Oceanic Technol., 19, 13181332, doi:10.1175/1520-0426(2002)019<1318:PAOTOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Buechler, D. E., Koshak W. J. , Chrisitan H. J. , and Goodman S. J. , 2014: Assessing the performance of the Lightning Imaging Sensor (LIS) using deep convective clouds. Atmos. Res.,135–136, 397403, doi:10.1016/j.atmosres.2012.09.008.

  • Christian, H. J., and Coauthors, 1999: The Lightning Imaging Sensor. 11th International Conference on Atmospheric Electricity, H. J. Christian, Ed., NASA Conf. Publ. NASA/CP—1999-209261, 746749. [Available online at http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19990108601.pdf.]

  • Dowden, R. L., Brundell J. B. , and Rodger C. J. , 2002: VLF lightning location by time of group arrival (TOGA) at multiple sites. J. Atmos. Sol.-Terr. Phys., 64, 817830, doi:10.1016/S1364-6826(02)00085-8.

    • Search Google Scholar
    • Export Citation
  • ENTLN, 2012: Total Lightning Network. [Available online at http://www.earthnetworks.com/Products/TotalLightningNetwork.aspx.]

  • Goodman, S. J., and Coauthors, 2005: The North Alabama Lightning Mapping Array: Recent severe storm observations and future prospects. Atmos. Res., 76, 423437, doi:10.1016/j.atmosres.2004.11.035.

    • Search Google Scholar
    • Export Citation
  • Goodman, S. J., and Coauthors, 2013: The GOES-R Geostationary Lightning Mapper (GLM). Atmos. Res., 125–126, 3449, doi:10.1016/j.atmosres.2013.01.006.

    • Search Google Scholar
    • Export Citation
  • Idone, V. P., and Orville R. E. , 1985: Correlated peak relative light intensity and peak current in triggered lightning subsequent return strokes. J. Geophys. Res., 90, 61596164, doi:10.1029/JD090iD04p06159.

    • Search Google Scholar
    • Export Citation
  • Jacobson, A. R., Holzworth R. , Harlin J. , Dowden R. , and Lay E. , 2006: Performance assessment of the World Wide Lightning Location Network (WWLLN), using the Los Alamos Sferic Array (LASA) as ground truth. J. Atmos. Oceanic Technol., 23, 10821092, doi:10.1175/JTECH1902.1.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., Krider P. , and Boccippio D. J. , 2000a: LIS validation at the KSC-CCAFS. Eos, Trans. Amer. Geophys. Union, 81 (Fall Meeting Suppl.), Abstract A52C-05, 2000.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., Stewart M. F. , Christian H. J. , Bergstrom J. W. , Hall J. M. , and Solakiewicz R. J. , 2000b: Laboratory calibration of the Optical Transient Detector and the Lightning Imaging Sensor. J. Atmos. Oceanic Technol., 17, 905915, doi:10.1175/1520-0426(2000)017<0905:LCOTOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Koshak, W. J., and Coauthors, 2004: North Alabama Lightning Mapping Array (LMA): VHF source retrieval algorithm and error analyses. J. Atmos. Oceanic Technol., 21, 543558, doi:10.1175/1520-0426(2004)021<0543:NALMAL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Krehbiel, P. R., Thomas R. J. , Rison W. , Hamlin T. , Harlin J. , and Davis M. , 2000: GPS-based mapping system reveals lightning inside storms. Eos, Trans. Amer. Geophys. Union, 81, 2125, doi:10.1029/00EO00014.

    • Search Google Scholar
    • Export Citation
  • Lay, E. H., Holzworth R. H. , Rodger C. J. , Thomas J. N. , Pinto O. Jr., and Dowden R. L. , 2004: WWLL global lightning detection system: Regional validation study in Brazil. Geophys. Res. Lett., 31, L03102, doi:10.1029/2003GL018882.

    • Search Google Scholar
    • Export Citation
  • Lay, E. H., Jacobson A. R. , Holzworth R. H. , Rodger C. J. , and Dowden R. L. , 2007: Local time variation in land/ocean lightning flash density as measured by the World Wide Lightning Location Network. J. Geophys. Res., 112, D13111, doi:10.1029/2006JD007944.

    • Search Google Scholar
    • Export Citation
  • Liu, C., and Heckman S. , 2010: The application of total lightning detection and cell tracking for severe weather prediction. TECO-2010-WMO Tech. Conf. on Meteorological and Environmental Instruments and Methods of Observation, Helsinki, Finland, WMO, P2(7). [Available online at https://www.wmo.int/pages/prog/www/IMOP/publications/IOM-104_TECO-2010/P2_7_Heckman_USA.pdf.]

  • Mach, D. M., Christian H. J. , Blakeslee R. J. , Boccippio D. J. , Goodman S. J. , and Boeck W. L. , 2007: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. J. Geophys. Res., 112, D09210, doi:10.1029/2006JD007787.

    • Search Google Scholar
    • Export Citation
  • Rison, W., Thomas R. J. , Krehbiel P. R. , Hamlin T. , and Harlin J. , 1999: A GPS-based three-dimensional lightning mapping system: Initial observations in central New Mexico. Geophys. Res. Lett., 26, 35733576, doi:10.1029/1999GL010856.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , Dowden R. L. , and Thomson N. R. , 2004: Location accuracy of long distance VLF lightning location network. Ann. Geophys., 22, 747758, doi:10.5194/angeo-22-747-2004.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , and Dowden R. L. , 2005: Location accuracy of VLF World-Wide Lightning Location (WWLL) network: Post-algorithm upgrade. Ann. Geophys., 23, 277290, doi:10.5194/angeo-23-277-2005.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Werner S. , Brundell J. B. , Lay E. H. , Thomson N. R. , Holzworth R. H. , and Dowden R. L. , 2006: Detection efficiency of the VLF World-Wide Lightning Location Network (WWLLN): Initial case study. Ann. Geophys., 24, 31973214, doi:10.5194/angeo-24-3197-2006.

    • Search Google Scholar
    • Export Citation
  • Rodger, C. J., Brundell J. B. , Holzworth R. H. , and Lay E. H. , 2009: Growing detection efficiency of the World Wide Lightning Location Network. AIP Conf. Proc.,1118, 15, doi:10.1063/1.3137706.

  • Rudlosky, S. D., and Shea D. T. , 2013: Evaluating WWLLN performance relative to TRMM/LIS. Geophys. Res. Lett., 40, 23442348, doi:10.1002/grl.50428.

    • Search Google Scholar
    • Export Citation
  • Taylor, W. L., 1972: Atmospherics and severe storms. Remote Sensing of the Troposphere, V. E. Derr, Ed., NOAA, 17-1–17-17.

  • Thomas, R. J., Krehbiel P. R. , Rison W. , Hamlin T. , Boccippio D. J. , Goodman S. J. , and Christian H. J. , 2000: Comparison of ground-based 3-dimensional lightning mapping observations with satellite-based LIS observations in Oklahoma. Geophys. Res. Lett., 27, 17031706, doi:10.1029/1999GL010845.

    • Search Google Scholar
    • Export Citation
  • Thomas, R. J., Krehbiel P. R. , Rison W. , Hunyady S. J. , Winn W. P. , Hamlin T. , and Harlin J. , 2004: Accuracy of the Lightning Mapping Array. J. Geophys. Res., 109, D14207, doi:10.1029/2004JD004549.

    • Search Google Scholar
    • Export Citation
  • Uman, M. A., 1987: The Lightning Discharge.Academic Press, 377 pp.

  • Ushio, T., Driscoll K. , Heckman S. , Boccippio D. , Koshak W. , and Christian H. , 1999: Initial comparison of the Lightning Imaging Sensor (LIS) with Lightning Detection and Ranging (LDAR). 11th International Conference on Atmospheric Electricity, H. J. Christian, Ed., NASA Conf. Publ. NASA/CP—1999-209261, 738–741. [Available online at http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19990108601.pdf.]

  • Fig. 1.

    Entire region 18-month (a) WWLLN and (b) ENTLN CP values based on constant, equal latitude and longitude constraints (line color) and a varying time constraint. The vertical black line runs through the chosen time constraint. (c),(d) As in (a),(b) but based on a constant time constraint (line color) and varying, equal latitude and longitude constraints. The vertical black line runs through the chosen latitude and longitude constraint. In all plots, the colored × symbols mark the tested constraints, and the black ♦ symbol marks the chosen coincidence criteria.

  • Fig. 2.

    Number of LIS groups in each 1° × 1° grid square for the 18 months. Note that the thick black lines mark the North American, South American, Atlantic Ocean, and Pacific Ocean subregion boundaries.

  • Fig. 3.

    The 3 × 3 smoothed 18-month WWLLN CP value of each 1° × 1° grid square.

  • Fig. 4.

    (bottom) The number of LIS groups in (left) the entire region and (right) North America, and time series plots for (top) WWLLN and (middle) ENTLN show the CP value each day from 1 Jan 2010 through 30 Jun 2011. Note that because LIS did not record any groups in North America on 4 Jan 2010, 19 Jan 2011, 9 Feb 2011, 11 Feb 2011, 22 Feb 2011, or 23 Feb 2011, those days do not have a CP value and resulted in a break in the line in the time series plots.

  • Fig. 5.

    The 3 × 3 smoothed 18-month ENTLN CP value of each 1° × 1° grid square.

  • Fig. 6.

    Number of LIS groups in each 1° × 1° grid square with at least one coincident ENTLN (a) ground stroke or (b) cloud stroke, and (c) GCP and (d) CCP values of each 1° × 1° grid square for the 18 months. Note that redundancy is included in these figures.

  • Fig. 7.

    Bar graphs that show monthly (a) WWLLN and (b) ENTLN CP values.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 1061 342 26
PDF Downloads 760 190 24