On-Orbit Response Characteristics of Fengyun-4A Lightning Mapping Imager (LMI) and Their Impacts on LMI Detection
1. Introduction
Effective monitoring of lightning provides valuable information to investigate the relationship between lightning and strong convective weather (Price 2000; Williams 2005; Siingh et al. 2013). Lightning now has been designated as an essential climate variable in the Global Climate Observing System to understand the climate change (Aich et al. 2018). Through the detection of electromagnetic fields and optical radiation generated by lightning, various lightning detection systems have been established on either ground-based or satellite-based platforms (Rison et al. 1999; Boccippio et al. 2001, 2002; Biagi et al. 2007; Liu and Heckman 2011; Hutchins et al. 2013; Nag et al. 2015; Rudlosky et al. 2017; Blakeslee et al. 2020b; F. Liu et al. 2021; Zhang et al. 2022). The ground-based lightning detection system has the ability to continuously monitor lightning within a global or regional range. The ground-based global system can detect lightning worldwide (including marine areas), such as the World Wide Lightning Location Network (WWLLN; Abarca et al. 2010; Hutchins et al. 2013) and Vaisala’s Global Lightning Dataset (GLD360; Said et al. 2013; Rudlosky et al. 2017). The ground-based regional system can detect regional lightning with a detection efficiency (DE) of 80%–90%, such as the Lightning detection Network (LINET; Betz et al. 2009), and the National Lightning Detection Network (NLDN) has a DE even higher than 90% (Nag et al. 2011, 2014; Zhu et al. 2016).
The satellite-based platform, e.g., the Optical Transient Detector (OTD) on the Microlab-1 satellite (Boccippio et al. 2000, 2002) and the Lightning Imaging Sensor (LIS) on the Tropical Rainfall Measuring Mission (TRMM) satellite (Christian et al. 2000; Albrecht et al. 2016), has entered a new era of total lightning (intracloud lightning and cloud-to-ground lightning) detection. Nowadays, the satellite-borne lightning detection is no longer limited to low-orbit satellite platforms. The geostationary satellite-borne lightning imager has extended the ability to track lightning and thunderstorms continuously over a large region (Christian et al. 1989; Chauzy et al. 2002). These instruments include the Geostationary Lightning Mappers (GLMs; Goodman et al. 2013; Rudlosky et al. 2019; Bateman et al. 2021; Rudlosky and Virts 2021) on board Geostationary Operational Environmental Satellite R-series (GOES-R) satellites, the Lightning Mapping Imager (LMI; Yang et al. 2017; Hui et al. 2020b) on board the Fengyun-4A (FY-4A) satellite, and the latest launched Lightning Imager (LI) on board the Meteosat Third Generation I (MTG-I) satellite (Kokou et al. 2018; EUMETSAT 2020, 2022).
The satellite-borne lightning imager operates in a narrow 1-nm spectral band centered on 777.4 nm (i.e., near-infrared; Christian et al. 2000) and detects the target relying on the observation of the radiation of light. On one hand, the detection capability of a lightning imager is related to the optical characteristics of the lightning source. The greater the radiant energy, the larger the area, and the longer the duration of the lightning flash, the more easily it tends to be detected (Marchand et al. 2019; Erdmann et al. 2020; Thiel et al. 2020; Zhang and Cummins 2020). Of course, the detection is also influenced by the optical thickness of the clouds, the viewing angle of the satellite, etc. (Finke 2009). On the other hand, the detection capability of the lightning imager is inseparable from its response characteristics (Zhang et al. 2019). Satellites observe the cloud-top radiation signals and identify the lightning signals based on the detector’s response. A pixel that exhibits an accumulated illumination above the background-adjusted threshold is designated a lightning event. Events occurring within the same frame and in adjacent pixels are clustered into a lightning group, and groups occurring within a specified time and distance are further clustered into a lightning flash (Mach et al. 2007). The response characteristics of the detector determine whether these lightning events can be detected and consequently whether the lightning flashes can be obtained (Christian et al. 2000). Zhang et al. (2019) found that the pixels in each of the four quadrants in the TRMM LIS detector have inconsistent response characteristics, resulting from different thresholds of the lightning detection in each quadrant. In the quadrant with the lowest threshold, the number of detected events increases by 15%–20% and the mean radiance of events decreases by 20% as compared to the other quadrants. In addition, the spatial resolution (pixel size) of the detector can also affect the observation of lightning sources. For example, the minimum detectable cloud-top energy of a GLM pixel is 2.58 times that of a TRMM LIS pixel. However, when the optical energy is spread uniformly over a GLM pixel (about 4 times larger than a TRMM LIS pixel; NASA 2019), a light source just detectable by the GLM will not be reported by the TRMM LIS because the energy within any of the LIS pixels is below its detection threshold (Zhang and Cummins 2020). Another example is the “pixel splitting” of the TRMM LIS due to its small pixels, which causes the radiance of the split light source to fail to reach the detection threshold, resulting in the miss of lightning detection (Zhang et al. 2019; Zhang and Cummins 2020).
At the end of 2016, China launched its first satellite-borne lightning imager, i.e., the LMI on board the FY-4A geostationary satellite (Yang et al. 2017). The LMI mission objectives are to continuously track and monitor lightning and severe weather in China and its surrounding areas (Liang et al. 2017). During more than 6 years’ in-orbit operation, the LMI has provided observational data for strong convection monitoring and numerical weather prediction in China (Liu et al. 2019; Y. Chen et al. 2020; Z. Chen et al. 2020; P. Liu et al. 2020; Zhang et al. 2020; Xian et al. 2021). Meanwhile, preliminary verifications and evaluations of its detection capability have been conducted (Hui et al. 2020b; R.-X. Liu et al. 2020; Cao et al. 2021; Chen et al. 2021; Hui and Guo 2021; Li et al. 2021; Y. Liu et al. 2021; Ni et al. 2021; Sun et al. 2021), and improvements on processing algorithms have been carried out.
Studies on LMI detection capability (R.-X. Liu et al. 2020; Cao et al. 2021; Hui and Guo 2021) have shown that the LMI has the ability to continuously observe lightning within a large field of view (FOV; Fig. 1). A comparative study showed that the spatial distribution of lightning density in China obtained by the LMI is consistent with the observation of the TRMM LIS and the WWLLN (Hui et al. 2020b), although some studies (Hui et al. 2020a,b; Y. Liu et al. 2021; Ni et al. 2021; Sun et al. 2021) indicated that the detection capability of the LMI varies greatly in different regions and exhibits diurnal variations. The number of lightning flashes observed by the LMI in some areas of western and central-eastern China is evidently lower than that observed by other instruments (Hui et al. 2020b). For example, the consistency of the detected lightning events over the Tibetan Plateau in western China between the LMI and the ISS LIS is only 3.66% (Hui et al. 2020a). In addition, the daytime/nighttime factor has an impact on LMI detection. The DE of the LMI in the daytime is at least 20% lower than that in the nighttime (Y. Liu et al. 2021; Sun et al. 2021). Ni et al. (2021) pointed out that during the daytime, the DE of the LMI in southern China was higher than that in other regions, and at night, the DE of the LMI was higher in central China and northeast China. To promote the operational application of LMI data, the reasons for the inconsistency of its detection capability need to be investigated.
Field of view (FOV) of the FY-4A LMI, including four regions [northwest (NW), blue contour; southwest (SW), cyan contour; northeast (NE), yellow contour; southeast (SE), red contour] and 16 subregions. The solid line indicates the boundary of the LMI FOV. The dashed line indicates the boundary of the subregions.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
The objective of this study is to examine the LMI detection capability from the perspective of the detector characteristics. In this study, by using the observation data of the FY-4A LMI in the Northern Hemisphere in 2020, and the overlapping observation data of the International Space Station (ISS) LIS with the LMI during the same period, we analyzed the response characteristics of the LMI detector and investigated the dependence of lightning properties (i.e., flash radiance, flash area, and flash duration) detected by LMI on its detector partitioning. Furthermore, the spatiotemporal changes in the LMI detection capability and their relationship with the response characteristics of LMI were discussed. The study will help improve the LMI calibration, event extracting algorithm, and application of the LMI observations.
2. Data and methods
a. FY-4A LMI data
The FY-4A LMI detects lightning flashes using spectral, spatial, and temporal filtering and background subtraction techniques (Bao et al. 2017; Yang et al. 2017). The LMI uses ultranarrow bandpass filters with a central wavelength of 777.4 nm and a bandwidth of 1 nm. It has a spatial resolution of 7.8 km at nadir and a frame rate of 2 ms. The background radiance of each frame is estimated based on the radiance of the current frame and its previous six frames (Bao et al. 2017). Through a frame-to-frame background subtraction, lightning events are extracted from the raw observation data. Details of the characteristics of LMI instrument and data processing algorithms are provided by Hui and Guo (2021).
The LMI provides a large FOV coverage based on dual lenses, and each lens focuses the image on a 400 × 300 pixel charge-coupled device (CCD) focal plane. Each focal plane is divided into eight subregions. Therefore, there are in total 16 subregions in the entire FOV of the LMI (Fig. 1). The output from each subregion is converted to digital information separately and then combined into a single data stream for further processing. The purpose of partition processing by the 16 subregions is to adapt to high-frame-rate sampling, and a similar approach is used in the GLM (NASA 2019; Rudlosky et al. 2019). According to the different occurrence probability of lightning flashes in different areas, the upper limit of the number of lightning events is set separately for each subregion. The upper limit of lightning events per frame is 120 (Liang et al. 2017). Although this number setting has ensured the downlinking of events as much as possible, the limitation on the number of lightning events for each subregion may still introduce some uncertainties to the analysis results of its response characteristics. As the FY-4A satellite platform flips up at the spring equinox and flips down at the autumn equinox, the LMI observes the Northern Hemisphere from mid-March to mid-September each year. For the remainder of the year, the LMI observes the Southern Hemisphere, covering western and central Australia and the neighboring waters (Hui et al. 2020a). Moreover, for a satellite-borne instrument, since the lightning illumination is detected above the cloud, a parallax issue will arise resulting in location uncertainty (Virts and Koshak 2020). Parallax refers to the fact that when the instrument observes the cloud top from a slantwise path, it will correspond to a different ground footprint from that observed in a nadir view (Wang et al. 2011). The potential for parallax errors increases with increasing incidence angle of the LMI and is largest at the northern edge of the LMI FOV.
Considering that more characteristics of a detector are reflected in its raw data, the LMI raw event data were used to analyze its response characteristics. The raw event data provide lightning event information on the time of occurrence, location, and digital number (DN) values, as well as DN values of the background and threshold corresponding to the event. The DN value is the digital measurement value obtained by the LMI detector and reflects the radiance of the target. The raw event data were converted from DN values to radiances using prelaunch calibration parameters of radiometric gain and offset. Here, the “radiance” is not really radiance, but rather a proxy to radiance, i.e., energy density (Koshak 2010). A lightning imager is designed to detect the transient lightning optical pulses from cloud top. The range of the background illumination affects the lightning detection threshold (Christian et al. 2000). It is complicated that the illumination conditions range from near 0 (such as night ocean as background) to peak energy (such as sun glint off the ocean), which means that the detector needs a large response range. In this study, the range of the event radiance from the minimum to the maximum is defined as dynamic range (Zhang et al. 2019), which reflects the response range of the detector to the lightning targets.
When analyzing lightning properties, the LMI L2 flash data were used. The data provide flash information on the time of occurrence, duration, location, radiance, area, and the numbers of groups and events clustering the flash. To improve the accuracy of the statistical results, the 3-sigma rule is used to define the outliers of the flash properties. The 3-sigma outliers were removed separately for each subregion at each hour, resulting in less than 3% of the total samples being removed for each property.
b. ISS LIS data
This study used lightning events and flashes data from the ISS LIS dataset to compare the observations between the LMI and the ISS LIS. The ISS LIS has the same detection method as the TRMM LIS and further expands the latitudinal coverage poleward to ±55°. It has a FOV of 78.5° × 78.5° based on a 128 × 128 CCD array and has a spatial resolution of approximately 4.5 km at nadir (Blakeslee and Koshak 2016; Blakeslee et al. 2018, 2020b). The flash DE of the ISS LIS is 64% relative to that of the GLM and 56%–57% relative to that of the Earth Networks Global Lightning Network (ENGLN) and the GLD360 with a subpixel location accuracy (<4 km; Blakeslee et al. 2020b).
The ISS LIS dataset includes science and background data of near–real time (NRT), non–quality controlled (NQC), and final quality controlled (QC). The NRT data are available within 2 min of observation, but only the most recent days are retained, and the historical data are deleted synchronously. The NQC data are created daily and the historical data are retained, but they are not subjected to specific quality control procedures. The QC data are gradually replacing NQC data through manual review to assure data quality (Blakeslee et al. 2020a). This study used the LIS QC event data to compare the event radiance observed simultaneously by LMI and LIS, and the QC flash data to compare the flash properties detected by LMI and LIS as well as to evaluate the detection capability of LMI. For each individual LIS event, the time of occurrence, location, radiance, and area are given. For each individual LIS flash, the time of occurrence, duration, location, radiance, area, and the number of inclusive events and groups are given. Considering the influence of viewing angles, orbit altitude, etc., on the pixel area of LIS, its event and flash areas were corrected by using Zhang et al.’s (2021) method.
c. Analysis methods
The lightning data used in this study are from 26 March to 20 September 2020, which was the observation period of the LMI for the Northern Hemisphere in 2020. ISS LIS data for the same period and area of LMI FOV were selected. The FOV is divided into four regions of the northwest (NW), the southwest (SW), the northeast (NE), and the southeast (SE). Each region is further divided into four subregions, e.g., NW1, NW2, NW3, and NW4. The subregions are delineated by the dashed lines in Fig. 1. In addition, to analyze the diurnal variation in the detection capability of the LMI, the periods of 0800–1600 local time (LT) and 1800–0600 LT are defined as daytime and nighttime, respectively.
(a) Coincidence ratio (CR) based on constant distance constraints (line color) and a varying time constraint. The chosen time constraint is marked by the dashed black line. (b) As in (a), but based on constant time constraints (line color) and a varying distance constraint. The chosen distance constraint is marked by the dashed black line. The black dot symbol marks the chosen coincidence criteria.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
A total of 135 945 140 LMI raw events and 50 153 LIS flashes within the LMI FOV in April and August were used as samples in the sensitivity test to determine the time/space coincidence window. For a LIS flash to be considered a match, the LMI event must occur within a designated distance from the LIS flash and within a designated time difference before, during, or after the LIS flash. Figure 2a shows the CR based on constant distance constraints and a varying time constraint. As the time difference increases, the CR increases. It can be seen that the curve takes relatively clear turns at 0.5, 1, and 1.5 s. After the time difference greater than 1.5 s, the increase of CR is slower relative to the previous increase (the dashed black line in Fig. 2a). Figure 2b shows the CR based on constant time constraints and a varying distance constraint. Due to the relatively large variation of CR with distance difference, and the location bias in LMI as indicated in previous studies (R.-X. Liu et al. 2020; Chen et al. 2021), the space coincidence window was tested to a maximum of 150 km. The CR tends to slow down when the distance difference is 60 km (the dashed black line in Fig. 2b). By trade-off of the above, a time/space coincidence window of 1.5 s/60 km (the black dot symbols in Fig. 2) was selected to compare the LMI with LIS observations. During 26 March to 20 September 2020, there were 114 740 LIS flashes that occurred within the LMI FOV. Based on the 1.5 s/60 km coincidence window, 28 197 LIS flashes are matched with LMI observations. The number of LIS flashes and matched LIS flashes in each subregion is summarized in Table 1.
Statistics of radiance and number of raw events detected by FY-4A LMI, number of ISS LIS flashes and the matched ISS LIS flashes, the parameters on LMI detection capability, and comparison of flash properties between the matched ISS LIS flashes and all ISS LIS flashes for the 16 subregions in the LMI field of view (FOV) during 26 Mar–20 Sep 2020. CR is the coincidence ratio between the LMI raw events and the LIS flashes.
3. Results and discussion
a. Response characteristics of the LMI detector
Figures 3a–e show the distribution of the minimum, maximum, 95% quantile, and mean radiance of the raw events detected by the LMI, and the number of raw events. Table 1 provides the corresponding statistical results. It is found that the optical sensitivity is inconsistent among subregions. The minimum event radiance of the southern regions (SW and SE) is generally lower than that of the northern regions (NW and NE; Fig. 3a). The lower the minimum event radiance, meaning a lower detection threshold (the distribution figure of the detection threshold is omitted, which is similar to Fig. 3a), the more events with weak radiance can be detected. The southern regions with 30% lower minimum event radiance are 22% lower mean event radiance (Fig. 3d; Table 1) and 20% more events (Fig. 3e; Table 1) compared to the northern regions. In contrast, in areas with higher minimum event radiance, there are more high radiance events in the total events. Figure 3f shows the mean event radiance detected by ISS LIS during the same period. In the four northernmost subregions (NW1, NW3, NE1, and NE3), the mean radiance of LMI events (Fig. 3d) is significantly higher than that of LIS events, which is related to the higher detection threshold of LMI in these subregions.
The distribution of the (a) minimum, (b) maximum, (c) 95% quantile, and (d) mean radiance of the raw events detected by FY-4A LMI, (e) the number of LMI raw events, and (f) the mean radiance of the ISS LIS events.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
It is interesting that the mean event radiance of subregions NW4, NE2, and NE4 is lower than that of the other regions, with that of subregion NW4 being particularly low. Correspondingly, there are more events in these three subregions than those in the surrounding area. Although the minimum event radiance of these three subregions is not obviously lower than that of the other regions, their dynamic range is 26% higher among all the regions (Table 1). A large dynamic range indicates the large span of radiance of the detectable events. Thus, the lightning frequency in the midlatitude region of China and its surrounding areas should be lower than that in the low-latitude region as shown in the studies of lightning climate (Cecil et al. 2011; Peterson et al. 2021), but the number of events detected by the LMI in subregions NW4, NE2, and NE4 is significantly higher than that in other regions at midlatitudes, and even the number of LMI events in subregion NW4 is higher than that in the low-latitude region (Fig. 3e; Table 1). Moreover, in these three subregions, the events with relatively low radiance compose a large fraction of the total events, especially subregion NW4 (Fig. 3c), which leads to the lower mean event radiance in these subregions compared to other regions.
Figure 4 shows the cumulative distribution function (CDF) of the maximum event radiance in the full focal plane array, the left focal plane (i.e., the western regions), and the right focal plane (i.e., the eastern regions). There is an obvious increase of the frequency of the full focal plane array near 2500 and 3300 μJ sr−1 m−2 nm−1, respectively (Fig. 4a), which is corresponding to the differences of the maximum radiance between the northern and the southern regions, as shown in Fig. 3b. The differences in the variation in the maximum radiance of each subregion (Figs. 4b,c) are due to their different nonlinear responses, and those of the southern regions are more consistent compared to the northern ones. Subregion SW3 in Fig. 4b is obviously different from other subregions in the SW region, showing a significant inconsistency in the response of pixels in this subregion. Moreover, the number of saturated pixels of the right focal plane is more than that of the left one, indicating that the right focal plane receives higher radiance. These all need to be considered for setting the detection threshold that varies with the background.
Cumulative distribution function (CDF) of the maximum event radiance in the (a) full focal plane array, (b) left focal plane, and (c) right focal plane.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
Figure 5 shows the diurnal variation in the mean radiance of the raw events in each subregion of the LMI. The event radiance of the southern regions shows a single peak variation, with the peak occurring at 1200–1400 LT, while some northern regions show a bimodal variation, with the main peak occurring still at 1200–1400 LT and the secondary peak at 2200–0300 LT, especially subregions NW1, NW2, NW3, NE1, and NE3. The main peak of the LMI event radiance occurs around noon, which indicates that the LMI event radiance is strongly influenced by the bright background. Not only that, there are obvious differences in the main radiance peaks among regions. For example, the main radiance peaks of the northern regions are higher than those of the southern ones, which is consistent with the difference in the maximum radiance between the northern and the southern regions (Fig. 4). This indicates that the daily peak of the LMI event radiance is also influenced by the inconsistency of the sensitivity among subregions.
Diurnal variation in the mean radiance of the raw events in each subregion of FY-4A LMI.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
The occurrence of the secondary radiance peak (approximately 2200–0300 LT) in the northern regions is due to stray light in the optical system of the LMI. The LMI instrument is installed on the satellite platform with a tilt to the north. The northern two corners of its FOV are close to the edge of Earth. When the satellite enters and exits the shadow of Earth, the sunlight shines directly into the optical system of the LMI. The stray light from multiple reflections of the sunlight in the optical system causes the background of LMI observations to be unusually bright. Thus, the events with unusually high radiance occur during this time period.
b. Response characteristics of the LMI detector on its observed flash properties
A flash detected by satellite can be regarded as a conventional lightning flash consisting of several strokes (Finke 2006). Figures 6a–f show the spatial distribution of flash properties observed by the LMI and the simultaneous observation results of the ISS LIS. The flash properties, which here mainly refer to the optical radiance, area, and duration of lightning flashes, can represent the intensity of lightning discharge (Williams et al. 2000; Beirle et al. 2014) and could also be related to the instrumental properties. Overall, the difference in values of all flash properties is obvious between the LMI and ISS LIS. Since the LMI detector’s response affects the lightning events it identifies, the flash properties associated with the events observed by the LMI could be affected by the inconsistency in the response of each subregion. The optical radiance of the LMI flashes (Fig. 6a) is generally higher than that of the LIS flashes (Fig. 6b). Previous studies have found that a large number of flashes with lower radiance can be detected by the LIS (Zhang et al. 2019; Hui et al. 2020b). The minimum detectable radiance of the LMI events (Fig. 3a) is significantly higher than that of the LIS events (Zhang et al. 2019), which contribute greatly to the higher radiance of the LMI flashes. The area of the LMI flashes (Fig. 6c) is generally greater than that of the LIS flashes (Fig. 6d). One possible reason is the lower spatial resolution of the LMI (7.8 km at nadir) compared to that of the LIS (4.5 km at nadir), indicating that the LMI observes a larger area illuminated by the lightning source. The LMI flash clustering process is also an influencing factor for its observed flash properties. The LMI uses the same flash clustering criterion within its FOV (section 2a), while the larger angle of incidence of the geostationary instrument causes parallax and larger pixels, which may cluster events that originally belong to one flash into more flashes, resulting in weaker and smaller flashes in regions with larger angle of incidence.
(a),(b) Optical radiance, (c),(d) area, and (e),(f) duration of lightning flashes, and (g),(h) number of groups in flashes observed by (left) FY-4A LMI and (right) ISS LIS.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
The distribution patterns of properties of the LMI and LIS flashes are compared. The most significant differences in flash radiance between the LMI and LIS are mainly concentrated in the northernmost subregions (NW1, NW3, NE1, and NE3) and the SW region (Figs. 6a,b). In these regions, a large number of flashes with high radiance are observed by the LMI, while only a small proportion of the flashes have high radiance observed by the LIS. As the LMI has higher minimum event radiance in the northern regions, followed by the SW region (Fig. 3a), it can detect lightning events with higher radiance in these regions and the radiance of the flashes clustered by these events remains higher. In addition, among these regions, subregions NW2, NW4, NE2, and NE4 have the larger dynamic range of event radiance (Table 1), and the events with relatively low radiance account for a large proportion of the total events (Fig. 3c), so that the radiance of LMI flashes obtained by clustering of the events in subregions NW2, NW4, NE2, and NE4 is relatively close to the radiance of LIS flashes. For the SE region, the flash radiance observed by the LMI is also relatively close to that observed by the LIS, which is due to the lower minimum event radiance in this region than in other regions (Fig. 3a).
Since the effect of the electrical and kinematic properties of thunderstorms on the charge distribution pattern (Bruning and MacGorman 2013; Zheng et al. 2016), the distribution of flash area tends to be similar to that of flash radiance, as observed by the LIS (Figs. 6b,d). However, there exists an inconsistency in the distribution of flash area and radiance observed by the LMI, which are mainly concentrated in the northernmost subregions and the SE region (Figs. 6a,c). The difference in the detectable event radiance of each subregion of the LMI is an important reason. For example, the SE region has the lowest minimum event radiance (Fig. 3a). The lower radiance events usually occur on the flash edge, leading the flash area to be enlarged (Zhang et al. 2019), so that the flashes in the SE region are larger than those in other regions. In contrast, in the northernmost subregions, the smaller flashes compared with other regions are influenced by the higher minimum event radiance in this region. Not only that, the larger LMI pixel size in the northernmost subregions is also a contributing factor. The pixels in the northernmost subregions can be 2–2.5 times larger than those in the southern regions within the LMI FOV. With the same flash clustering criterion, the larger the pixels, the events that originally belong to the same flash are clustered into multiple flashes, resulting in smaller flashes.
The flash duration is related to the number of its component strokes, i.e., the more the component strokes, the longer the flash duration, but there is not definite relationship between the two (Malan 1956). Considering that a group can be regarded as a cloud-to-ground (CG) return stroke or an intracloud (IC) discharge to some extent (Marchand et al. 2019), Figs. 6g and 6h show the number of groups in flashes observed by the LMI and the LIS, respectively. A comparison of Figs. 6f and 6h finds that the duration of LIS flashes and the number of its component groups are similar in spatial distribution. The distribution of LMI flash durations and the number of groups they include (Figs. 6e,g) are also generally consistent, except for obvious differences in subregions NW4, NE2, and NE4. Based on the TRMM LIS observations over a 14-yr period, Peterson et al. (2017) investigated the flash properties measured by the TRMM LIS. The variation in the flash properties results from the energetics and structure of the lightning flashes as well as radiative transfer within the cloud. According to their results, the flash duration and the number of groups per flash show the highest positive correlation, with a correlation coefficient of 0.7, which is in accordance with the high consistency of the two distributions observed by the ISS LIS and the LMI in this study. For the exceptions in subregions NW4, NE2, and NE4, most LMI flashes have relatively longer durations. Nevertheless, the component groups per flash in this area are not more than other regions. This may be attributed to the larger dynamic range of event radiance (Table 1) and the low radiance of most events in this area (Fig. 3c). In addition, the LMI flashes in subregion NW2 show longer duration and more groups per flash, which may be related to the applicability of the LMI flash clustering criterion to different regions. Subregion NW2 and its surrounding areas correspond to the Tibetan Plateau, where there are unique thermal and dynamical effects, and thus often experience high frequency but relatively weak convective activities (Gao et al. 1981; Qie et al. 2003; Iwasaki 2016). The flashes on the Plateau consist of only very few groups and last shorter as observed by the LIS (Figs. 6f,h). However, with the current spatial and temporal clustering criterion, the LMI groups in this region that originally belong to different flashes may be clustered into one flash, resulting in longer flash duration and more groups per flash.
Figure 7 shows the diurnal variations in the mean optical radiance, area, and duration of flashes, the number of groups in flashes, and the number of flashes in each subregion observed by the LMI, and also compared with the LIS observations. The radiance of the LMI flashes in each subregion exhibits a single peak diurnal variation (Fig. 7a), with the peak occurring around noon. The LMI flash radiance during the day is significantly higher than that at night. The diurnal variation in the LMI flash radiance is similar to that in the mean radiance of the raw events observed by the LMI (Fig. 5). In fact, there is no obvious difference in flash radiance between daytime and nighttime, as shown by the LIS observation (Fig. 7b), which can also be illustrated by You et al.’s (2019) study. The peak of the LMI flash radiance mainly indicates that its event extraction is influenced by the bright background, and may be related to the number of flashes after clustering. As shown in Fig. 7i, the LMI flashes around noon are much less than those during other hours of the day. Thus, their statistical results represent only a limited number of thunderstorm processes. For example, the extreme values of flash radiance of subregion SW3 correspond to a few strong thunderstorms that occurred in this region in mid-April and late May, respectively. Moreover, there are regional differences in the diurnal variation in the flash radiance due to the inconsistency of the sensitivity among subregions. However, the regional differences are not exactly consistent with those of the diurnal variation in the event radiance (Fig. 5). Specifically, the peaks of LMI flash radiance of the southern regions are higher than those of the northern ones, while the peaks of LMI event radiance of the southern regions are lower than those of the northern ones. The events with higher peak radiance in the northern regions are clustered, but the resulting flash peak radiance is weaker. The larger pixels in the northern regions, when using the same flash clustering criterion as in the southern regions, may cause the events that originally belong to the same flash are clustered into multiple flashes, and thus, the flash radiance becomes weaker. The limited sample size may also be an influencing factor.
Diurnal variations in the mean (a),(b) optical radiance, (c),(d) area, and (e),(f) duration of (left) LMI and (right) LIS flashes. As in (a)–(f), but for the number of groups in (g) LMI and (h) LIS flashes, and number of (i) LMI and (j) LIS flashes in each subregion.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
The diurnal pattern of the LMI-detected flash area in the southern regions shows a single peak variation with the peak occurring around noon, while the flash area in the northern regions shows a relatively flat variation with no evident peak (Fig. 7c). Considering the relationship between flash radiance and area (Bruning and MacGorman 2013; Zheng and MacGorman 2016; Peterson et al. 2017), the diurnal variations of the two show similar trend, as observed by the LIS (Figs. 7b,d). Meanwhile, there is no obvious difference in the LIS flash area between daytime and nighttime, like the LIS flash radiance. However, the LMI observes the peak in flash area in some regions around noon, again indicating the influence of bright background on the LMI event extraction, as well as the influence of the lower number of flashes around noon. Moreover, the flash in the northern regions is about 1/3 smaller than that in the southern ones, which is still related to the effect of pixel size on LMI flash clustering.
The duration of LMI flashes and the number of its component groups in each subregion show similar diurnal variation trends (Figs. 7e,g), just as the two have a high similarity in spatial distribution. There are differences in the diurnal variation between flash duration and the number of groups per flashes observed by the LIS, especially in the first half of the day (Figs. 7f,h). The significantly lower number of LIS flashes during this time period should be a contributing factor (Fig. 7j), which also demonstrates the contribution of sample size. In addition, some peaks of the LMI flash duration and the number of groups in LMI flashes are related to the sample size, resulting in the statistical results that are affected by a limited number of thunderstorm processes. For example, the peak of subregion NE1 at 0300 LT mainly corresponds to a thunderstorm process that occurred near 56°N, 124°E on 3 July 2020, during which the mean duration was 828.54 ms and the mean number of groups was 23.
c. Response characteristics of the LMI detector on its detection capability
Previous studies have found that the detection capability of the LMI varies greatly in different regions and exhibits diurnal variations (Hui et al. 2020a,b; Y. Liu et al. 2021; Ni et al. 2021; Sun et al. 2021). The CR is defined in this study to quantitatively analyze the detection capability of the LMI relative to the ISS LIS. This comparative approach was also employed by the studies of the detection capability of the GLM (Marchand et al. 2019; Murphy and Said 2020; Bateman et al. 2021). Figure 8 shows the CR between the LMI and LIS within the FOV of the LMI, and Table 1 gives the mean CR values in each subregion. Figure 8 shows that the overall CR in the southern regions is higher than that in the northern regions and higher in the eastern regions than in the western regions. The mean CR values of the SW region and the SE region are 31.02% and 28.39%, respectively, followed by the mean CR of 25.52% in the NE region and the lowest of 7.11% in the NW region (Table 1). The heterogeneous spatial distribution of CR not only confirms the different detection capabilities of each subregion of the LMI detector but also shows that the detection capability of the right focal plane of the LMI is higher than that of the left one.
The spatial distribution of CR between FY-4A LMI and ISS LIS.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
The inconsistency in CR is related to the difference in the response of each subregion of the LMI. Figure 9 shows the CDF of the radiance of the LMI raw events in each subregion. The radiance of events in the southern regions is overall low, indicating that weaker lightning signals can be detected in this region. Correspondingly, the CR in the southern regions is relatively high. The radiance of events in the NE region is not overall low, but its dynamic range is higher than that in the southern ones, making the CR in this region comparable to that in the southern ones. In the NW region, especially subregions NW1, NW2, and NW3, the event radiance is obviously higher than that in other regions, and their dynamic ranges are relatively low, which affects the detection capability in this region and the CR is the lowest. The relatively high radiance in subregions NW1, NW2, and NW3 may be related to the unusual sunlight intrusion around midnight, and the same is the case for subregions NE1 and NE3. Moreover, subregions NW4, NE2, and NE4 still show interesting characteristics. The event radiance in these three subregions is relatively low and, in particular, their dynamic ranges are significantly higher than other regions, which is conducive to the identification of targets with a large variation range of radiance. As a result, these three subregions have a high CR in the northern regions.
CDF of the radiance of the raw events in each subregion of FY-4A LMI.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
Moreover, considering that the CR shows the LMI detection relative to the LIS, the LIS characteristics also influence the CR. The ratio of flash properties of the matched LIS flashes to the total LIS flashes was computed and their distribution is illustrated in Fig. 10. A statistical comparison of these properties for each subregion is summarized in Table 1. The high values of the ratio of all flash properties are concentrated in the SW, NE, and SE regions, which is coincident with the distribution of CR values (Fig. 8). Considering the flash radiance, for example, the mean ratio of matched flash radiance to total flash radiance in the NW region is 0.49, which is only about 75% of the other regions (Table 1). This indicates that compared to the NW region, the radiance of matched flashes is obviously higher than the mean radiance of all flashes in the other regions. Correspondingly, the mean CR of the NW region is only 32% of that of the other ones. Therefore, it can be concluded that the LMI tends to detect LIS flashes with high radiance, large areas, and long duration.
The spatial distribution of the (a) optical radiance, (b) area, and (c) duration ratio between the matched LIS flashes and all LIS flashes.
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
Figure 11 shows the mean CR of each subregion during the daytime (0800–1600 LT) and nighttime (1800–0600 LT). The mean CR of each subregion is 8.82% during the day and 31.48% at night (Table 1). The background radiance in the daytime is higher, so that the dynamic detection threshold should become higher. However, the low overall CR during the day indicates that the detection threshold of LMI is higher than expected, resulting in its extraction of fewer events. Moreover, the differences in detection capability between the northern and southern regions are more obvious during the day compared to the night. During the daytime, the mean CR is 2.44 times higher in the southern regions than in the northern ones and 1.55 times higher at night. This suggests that the differences in sensitivity among subregions have a greater impact on the LMI detection capability during the day, all of which need to be considered for improvement in detection threshold in future.
Mean CR of each subregion during the daytime (0800–1600 LT) and nighttime (1800–0600 LT).
Citation: Journal of Atmospheric and Oceanic Technology 40, 12; 10.1175/JTECH-D-22-0126.1
4. Conclusions
In this paper, the detection capability of the LMI was shown to have regional and diurnal differences in the application of the FY-4A LMI data. The effect of the partition design of the LMI detector on the observation results was studied. The response characteristics of the LMI detector to the lightning source were analyzed based on the LMI raw event data in the Northern Hemisphere in 2020, and the relationship between the response characteristics of the LMI detector and the flash properties observed by LMI was studied by taking advantage of the simultaneous observation data of the ISS LIS. Then, the spatiotemporal changes in the detection capability of the LMI due to the response characteristics of the LMI detector were investigated. The main conclusions obtained are as follows:
-
The difference in sensitivity of each subregion of LMI detector to the lightning source results in spatial difference in the number of detectable events. Among the 16 subregions of the LMI FOV analyzed, the minimum event radiance of the southern regions is generally lower than that of the northern ones, meaning a lower detection threshold in the southern regions. Therefore, more events with weak radiance can be detected in the southern regions, which is conducive to the identification of events. The minimum event radiance of subregions NW4, NE2, and NE4 is not obviously lower than that of the other regions, but their dynamic range is 26% higher among all the regions, which leads to more events in these three subregions being detected. Moreover, the saturated value of each subregion is different due to their different nonlinear response as a function of the background level, and the right focal plane tends to receive higher radiance than the left one, which needs to be considered for setting the dynamic detection threshold.
-
The mean radiance of the LMI raw events has obvious diurnal variation characteristics, and the diurnal variation characteristics differ among different subregions. The event radiance of the southern regions shows a single peak variation, with the peak occurring at 1200–1400 LT, while that of the northern regions shows a bimodal variation, with the main peak occurring still at 1200–1400 LT and the secondary peak at 2200–0300 LT. The main peak around noon indicates that the LMI event radiance is strongly influenced by the bright background, and the difference in the main peaks of different subregions is consistent with the difference in the maximum event radiance, which demonstrates the effect of inconsistency of the sensitivity among subregions. The stray sunlight when the satellite enters and exits Earth’s shadow around midnight results in the secondary radiance peak in the northern regions.
-
The response capability of the LMI detector affects the flash properties detected by the LMI. Overall, the radiance of LMI flashes is higher than that of LIS, which is due to the higher minimum detectable radiance of LMI events. Moreover, the LMI-detected flash properties show patterns related to the inconsistency of the sensitivity among subregions. Specifically, the proportion of flashes with high radiance in the northernmost subregions and the SW region detected by the LMI is significantly greater than that detected by the LIS; the distribution of flash area tends to be similar to that of flash radiance, as observed by the LIS, but the two differ in the northernmost subregions and the SE region as observed by the LMI; the duration of flashes and the number of its component groups observed by the LIS are similar, while obvious differences of the two in subregions NW4, NE2, and NE4 are observed by the LMI. In terms of temporal variation, the flash radiance and area observed by the LMI exhibit a single peak around noon, which is mainly influenced by the bright background.
-
The difference in the response capability of each subregion of LMI affects its detection, so that the CR shows large spatial variability. The SW region has the highest CR, followed by the SE region and the NE region, and the NW region has the lowest CR. In other words, the detection capability of the southern regions is higher than that of the northern regions, and that in the right focal plane of LMI is higher than in the left one. This inconsistency can be explained by the radiance and the dynamic range of the detectable events in each subregion. Furthermore, a comparison of the flash properties of the matched and total LIS flashes demonstrates that the LMI tends to detect LIS flashes with high radiance, large areas, and long duration. As for the diurnal variation in the CR, the CR at night is higher than that in the day. The detection threshold of LMI should be higher in the day, but it may be higher than expected, resulting in fewer events being extracted by LMI during this time period. Meanwhile, during the daytime, the differences in sensitivity among regions have a greater impact on the LMI detection capability compared to the nighttime.
Due to the limitation of detector development technology, the LMI detector has its own characteristics. Based on the raw event data of the LMI, this study investigated the differences in the responses of different subregions of the LMI detector and their effect on the LMI detection capability. On one hand, this study can help data users better understand the current observation results of the LMI and more effectively apply the data. According to the results of this study, users can understand the reasons behind a low CR in some regions and time periods and can use the data of appropriate region and time based on the regional changes and diurnal variations in CR. On the other hand, according to the characteristics of pixel energy distribution, algorithm developers can improve the LMI calibration and event extracting algorithm. By adjusting the observation inconsistency caused by the different responses of different subregions, the detection capability of LMI can be improved.
Acknowledgments.
This work was supported by the National Key Research and Development Program of China (2019YFC1510103), the National Natural Science Foundation of China (42205138), the Open Grants of the State Key Laboratory of Severe Weather (2022LASW-B12), and the Basic Research Fund of CAMS (2020Z009). The authors wish to thank the NASA Global Hydrology Resource Center (GHRC) Data Analysis and Archive Center (DAAC) for providing the science data of the ISS LIS. The FY-4A LMI data were provided by the National Satellite Meteorological Center (NSMC), China Meteorological Administration (CMA). The authors thank the anonymous reviewers for their valuable comments that considerably improved the paper, along with Dr. Daile Zhang for her assistance in correcting the LIS data.
Data availability statement.
The science data of the ISS LIS are available from the NASA GHRC DAAC (https://search.earthdata.nasa.gov/portal/ghrc/search?fi=LIS). The FY-4A LMI data can be downloaded from the Fengyun Satellite Data Center (http://satellite.nsmc.org.cn/PortalSite/Default.aspx?currentculture=en-US) or requested offline via email (dataserver@cma.gov.cn). Meanwhile, the data supporting the conclusions of this study are available at https://doi.org/10.5281/zenodo.8146694.
REFERENCES
Abarca, S. F., K. L. Corbosiero, and T. J. Galarneau 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, https://doi.org/10.1029/2009JD013411.
Aich, V., R. Holzworth, S. Goodman, Y. Kuleshov, C. Price, and E. Williams, 2018: Lightning: A new essential climate variable. Eos, 99, https://doi.org/10.1029/2018EO104583.
Albrecht, R. I., S. J. Goodman, D. E. Buechler, R. J. Blakeslee, and H. J. Christian, 2016: Where are the lightning hotspots on Earth? Bull. Amer. Meteor. Soc., 97, 2051–2068, https://doi.org/10.1175/BAMS-D-14-00193.1.
Bao, S., Y. Li, S. Tang, H. Liang, and X. Zhao, 2017: Instantaneous real-time detection technology of GLI on FY-4 geostationary meteorological satellite. Aerosp. China, 18, 23–30.
Bateman, M., D. Mach, and M. Stock, 2021: Further investigation into detection efficiency and false alarm rate for the Geostationary Lightning Mappers aboard GOES-16 and GOES-17. Earth Space Sci., 8, e2020EA001237, https://doi.org/10.1029/2020EA001237.
Beirle, S., W. Koshak, R. Blakeslee, and T. Wagner, 2014: Global patterns of lightning properties derived by OTD and LIS. Nat. Hazards Earth Syst. Sci., 14, 2715–2726, https://doi.org/10.5194/nhess-14-2715-2014.
Betz, H. D., K. Schmidt, P. Laroche, P. Blanchet, W. P. Oettinger, E. Defer, Z. Dziewit, and J. Konarski, 2009: LINET—An international lightning detection network in Europe. Atmos. Res., 91, 564–573, https://doi.org/10.1016/j.atmosres.2008.06.012.
Biagi, C., K. Cummins, K. Kehoe, and E. Krider, 2007: National Lightning Detection Network (NLDN) performance in southern Arizona, Texas, and Oklahoma in 2003–2004. J. Geophys. Res., 112, D05208, https://doi.org/10.1029/2006JD007341.
Blakeslee, R. J., and W. Koshak, 2016: LIS on ISS: Expanded global coverage and enhanced applications. Earth Observer, Vol. 28, No. 3, NASA Science Support Office, Greenbelt, MD, 4–14.
Blakeslee, R. J., and Coauthors, 2018: Lightning Imaging Sensor on the International Space Station: Assessments and results from first year operations. Proc. 16th Int. Conf. on Atmospheric Electricity, Nara, Japan, ICAE.
Blakeslee, R. J., D. M. Mach, and M. F. Stewart, 2020a: International Space Station (ISS) Lightning Imaging Sensor (LIS) data user guide. NASA GHRC Doc., 24 pp.
Blakeslee, R. J., and Coauthors, 2020b: Three years of the Lightning Imaging Sensor onboard the International Space Station: Expanded global coverage and enhanced applications. J. Geophys. Res. Atmos., 125, e2020JD032918, https://doi.org/10.1029/2020JD032918.
Boccippio, D. J., and Coauthors, 2000: The Optical Transient Detector (OTD): Instrument characteristics and cross-sensor validation. J. Atmos. Oceanic Technol., 17, 441–458, https://doi.org/10.1175/1520-0426(2000)017<0441:TOTDOI>2.0.CO;2.
Boccippio, D. J., S. Heckman, and S. J. Goodman, 2001: A diagnostic analysis of the Kennedy Space Center LDAR network: 1. Data characteristics. J. Geophys. Res., 106, 4769–4786, https://doi.org/10.1029/2000JD900687.
Boccippio, D. J., W. J. Koshak, and R. J. Blakeslee, 2002: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. Part I: Predicted diurnal variability. J. Atmos. Oceanic Technol., 19, 1318–1332, https://doi.org/10.1175/1520-0426(2002)019<1318:PAOTOT>2.0.CO;2.
Bruning, E. C., and D. R. MacGorman, 2013: Theory and observations of controls on lightning flash size spectra. J. Atmos. Sci., 70, 4012–4029, https://doi.org/10.1175/JAS-D-12-0289.1.
Cao, D., F. Lu, X. Zhang, and J. Yang, 2021: Lightning activity observed by the Fengyun-4A Lightning Mapping Imager. Remote Sens., 13, 3013, https://doi.org/10.3390/rs13153013.
Cecil, D. J., D. E. Buechler, and R. J. Blakeslee, 2011: TRMM-based lightning climatology. Proc. 14th Int. Conf. on Atmospheric Electricity, Rio de Janeiro, Brazil, ICAE.
Chauzy, S., S. Coquillat, and S. Soula, 2002: On the relevance of lightning imagery from geostationary satellite observation for operational meteorological applications. EUMETSAT Tech. Rep. EUM/COL/LET/02/1562, 76 pp.
Chen, Y., Z. Yu, W. Han, J. He, and M. Chen, 2020: Case study of a retrieval method of 3D proxy reflectivity from FY-4A lightning data and its impact on the assimilation and forecasting for severe rainfall storms. Remote Sens., 12, 1165, https://doi.org/10.3390/rs12071165.
Chen, Z., J. Sun, X. Qie, Y. Zhang, Z. Ying, X. Xiao, and D. Cao, 2020: A method to update model kinematic states by assimilating satellite-observed total lightning data to improve convective analysis and forecasting. J. Geophys. Res. Atmos., 125, e2020JD033330, https://doi.org/10.1029/2020JD033330.
Chen, Z., X. Qie, J. Sun, X. Xiao, Y. Zhang, D. Cao, and J. Yang, 2021: Evaluation of Fengyun-4A Lightning Mapping Imager (LMI) performance during multiple convective episodes over Beijing. Remote Sens., 13, 1746, https://doi.org/10.3390/rs13091746.
Christian, H. J., R. J. Blakeslee, and S. J. Goodman, 1989: The detection of lightning from geostationary orbit. J. Geophys. Res., 94, 13 329–13 337, https://doi.org/10.1029/JD094iD11p13329.
Christian, H. J., R. J. Blakeslee, S. J. Goodman, and D. M. Mach, 2000: Algorithm theoretical basis document for the Lightning Imaging Sensor (LIS). NASA Measures Algorithm Theoretical Basis Doc., 53 pp., https://eospso.gsfc.nasa.gov/sites/default/files/atbd/atbd-lis-01.pdf.
Erdmann, F., E. Defer, O. Caumont, R. J. Blakeslee, S. Pédeboy, and S. Coquillat, 2020: Concurrent satellite and ground-based lightning observations from the Optical Lightning Imaging Sensor (ISS-LIS), the low-frequency network Meteorage and the SAETTA Lightning Mapping Array (LMA) in the northwestern Mediterranean region. Atmos. Meas. Tech., 13, 853–875, https://doi.org/10.5194/amt-13-853-2020.
EUMETSAT, 2020: Meteosat Third Generation Lightning Imager level 2 expected performances. EUMETSAT Tech. Rep., EUM/RSP/REP/20/1179001, 44 pp., https://www-cdn.eumetsat.int/files/2020-12/Meteosat%20Third%20Generation%20Lightning%20Imager%20Level%202%20expected%20performances_web.pdf.
EUMETSAT, 2022: LI L2 product user guide. EUMETSAT Tech. Rep., EUM/GEO/TEN/15/828715, 168 pp., https://www.eumetsat.int/media/49348.
Finke, U., 2006: Characterising the lightning source for the MTG Lightning Imager mission. EUMETSAT Tech. Rep., EUM/MTG/SOW/05/0028, 46 pp.
Finke, U., 2009: Optical detection of lightning from space. Lightning: Principles, Instruments and Applications, H. D. Betz, U. Schumann, and P. Laroche, Eds., Springer Science and Business Media, 271–286.
Gao, Y., M. Tang, S. Luo, Z. Shen, and C. Li, 1981: Some aspects of recent research on the Qinghai-Xizang Plateau meteorology. Bull. Amer. Meteor. Soc., 62, 31–35, https://doi.org/10.1175/1520-0477(1981)062<0031:SAORRO>2.0.CO;2.
Goodman, S. J., and Coauthors, 2013: The GOES-R Geostationary Lightning Mapper (GLM). Atmos. Res., 125–126, 34–49, https://doi.org/10.1016/j.atmosres.2013.01.006.
Hui, W., and Q. Guo, 2021: Preliminary characteristics of measurements from Fengyun-4A Lightning Mapping Imager. Int. J. Remote Sens., 42, 4922–4941, https://doi.org/10.1080/01431161.2021.1906983.
Hui, W., F. Huang, and R. Liu, 2020a: Characteristics of lightning signals over the Tibetan Plateau and the capability of FY-4A LMI lightning detection in the plateau. Int. J. Remote Sens., 41, 4605–4625, https://doi.org/10.1080/01431161.2020.1723176.
Hui, W., W. Zhang, W. Lyu, and P. Li, 2020b: Preliminary observations from the China Fengyun-4A Lightning Mapping Imager and its optical radiation characteristics. Remote Sens., 12, 2622, https://doi.org/10.3390/rs12162622.
Hutchins, M. L., R. H. Holzworth, K. S. Virts, J. M. Wallace, and S. Heckman, 2013: Radiated VLF energy differences of land and oceanic lightning. Geophys. Res. Lett., 40, 2390–2394, https://doi.org/10.1002/grl.50406.
Iwasaki, H., 2016: Relating lightning features and topography over the Tibetan Plateau using the World Wide Lightning Location Network data. J. Meteor. Soc. Japan, 94, 431–442, https://doi.org/10.2151/jmsj.2016-025.
Kokou, P., P. Willemsen, M. Lekouara, M. Arioua, A. Mora, P. V. Braembussche, E. Neri, and D. M. A. Aminou, 2018: Algorithmic chain for lightning detection and false event filtering based on the MTG Lightning Imager. IEEE Trans. Geosci. Remote Sens., 56, 5115–5124, https://doi.org/10.1109/TGRS.2018.2808965.
Koshak, W. J., 2010: Optical characteristics of OTD flashes and the implications for flash-type discrimination. J. Atmos. Oceanic Technol., 27, 1822–1838, https://doi.org/10.1175/2010JTECHA1405.1.
Koshak, W. J., 2017: Lightning NOx estimates from space-based lightning imagers. 16th Annual Community Modeling and Analysis System Conf., Chapel Hill, NC, University of North Carolina at Chapel Hill.
Li, P., G. Zhai, W. Pang, W. Hui, W. Zhang, J. Chen, and L. Zhang, 2021: Preliminary research on a comparison and evaluation of FY-4A LMI and ADTD data through a moving amplification matching algorithm. Remote Sens., 13, 11, https://doi.org/10.3390/rs13010011.
Liang, H., S. Bao, Q. Chen, X. Zhao, and Y. Li, 2017: Design and implementation of FY-4 Geostationary Lightning Imager. Aerosp. Shanghai, 34, 43–51.
Liu, C., and S. Heckman, 2011: Using total lightning data in severe storm prediction: Global case study analysis from North America, Brazil and Australia. 2011 Int. Symp. on Lightning Protection, Fortaleza, Brazil, IEEE, 20–24, https://doi.org/10.1109/SIPDA.2011.6088433.
Liu, F., and Coauthors, 2021: Meteorological and electrical conditions of two mid-latitude thunderstorms producing blue discharges. J. Geophys. Res. Atmos., 126, e2020JD033648, https://doi.org/10.1029/2020JD033648.
Liu, P., Y. Yang, J. Gao, Y. Wang, and C. Wang, 2020: An approach for assimilating FY4 lightning and cloud top height data using 3DVAR. Front. Earth Sci., 8, 288, https://doi.org/10.3389/feart.2020.00288.
Liu, R.-X., T. Liu, A. Pessi, W. Hui, W. Cheng, and F.-X. Huang, 2019: Preliminary study on the influence of FY-4 lightning data assimilation on precipitation predictions. J. Trop. Meteor., 25, 528–541.
Liu, R.-X., Q.-F. Lu, M. Chen, Y. Zhang, W. Hui, and X.-Q. Li, 2020: Quality assessment of FY-4A lightning data in inland China. J. Trop. Meteor., 26, 286–299.
Liu, Y., H. Wang, Z. Li, and Z. Wang, 2021: A verification of the lightning detection data from FY-4A LMI as compared with ADTD-2. Atmos. Res., 248, 105163, https://doi.org/10.1016/j.atmosres.2020.105163.
Mach, D. M., H. J. Christian, R. J. Blakeslee, D. J. Boccippio, S. J. Goodman, and W. L. Boeck, 2007: Performance assessment of the Optical Transient Detector and Lightning Imaging Sensor. J. Geophys. Res., 112, D09210, https://doi.org/10.1029/2006JD007787.
Malan, D. J., 1956: The relation between the number of strokes, stroke intervals and the total durations of lightning discharges. Pure Appl. Geophys., 34, 224–230, https://doi.org/10.1007/BF02122829.
Marchand, M., K. Hilburn, and S. D. Miller, 2019: Geostationary Lightning Mapper and Earth Networks lightning detection over the contiguous United States and dependence on flash characteristics. J. Geophys. Res. Atmos., 124, 11 552–11 567, https://doi.org/10.1029/2019JD031039.
Murphy, M. J., and R. K. Said, 2020: Comparisons of lightning rates and properties from the U.S. National Lightning Detection Network (NLDN) and GLD360 with GOES-16 Geostationary Lightning Mapper and Advanced Baseline Imager data. J. Geophys. Res. Atmos., 125, e2019JD031172, https://doi.org/10.1029/2019JD031172.
Nag, A., and Coauthors, 2011: Evaluation of U.S. National Lightning Detection Network performance characteristics using rocket-triggered lightning data acquired in 2004–2009. J. Geophys. Res., 116, D02123, https://doi.org/10.1029/2010JD014929.
Nag, A., M. J. Murphy, K. L. Cummins, A. E. Pifer, and J. A. Cramer, 2014: Recent evolution of the U.S. National Lightning Detection Network. Proc. 23rd Int. Lightning Detection Conf., Tucson, AZ, Vaisala, https://www.vaisala.com/sites/default/files/documents/Nag%20et%20al-Recent%20Evolution%20of%20the%20U.S.%20National%20Lightning%20Detection%20Network-2014-ILDC-ILMC.pdf.
Nag, A., M. J. Murphy, W. Schulz, and K. L. Cummins, 2015: Lightning locating systems: Insights on characteristics and validation techniques. Earth Space Sci., 2, 65–93, https://doi.org/10.1002/2014EA000051.
NASA, 2019: GOES-R series data book. NASA Tech. Rep. CDRL PM-14, 240 pp., https://www.goes-r.gov/downloads/resources/documents/GOES-RSeriesDataBook.pdf.
Ni, X., W. Hui, Q. Zhang, F. Huang, and C. Liu, 2021: Comparison of lightning detection between the FY-4A Lightning Mapping Imager and the ISS Lightning Imaging Sensor. Earth Space Sci., 8, e2020EA001099, https://doi.org/10.1029/2020EA001099.
Peterson, M., W. Deierling, C. Liu, D. Mach, and C. Kalb, 2017: The properties of optical lightning flashes and the clouds they illuminate. J. Geophys. Res. Atmos., 122, 423–442, https://doi.org/10.1002/2016JD025312.
Peterson, M., D. Mach, and D. Buechler, 2021: A global LIS/OTD climatology of lightning flash extent density. J. Geophys. Res. Atmos., 126, e2020JD033885, https://doi.org/10.1029/2020JD033885.
Price, C., 2000: Evidence for a link between global lightning activity and upper tropospheric water vapour. Nature, 406, 290–293, https://doi.org/10.1038/35018543.
Qie, X., R. Toumi, and Y. Zhou, 2003: Lightning activity on the central Tibetan Plateau and its response to convective available potential energy. Chin. Sci. Bull., 48, 296–299, https://doi.org/10.1007/BF03183302.
Rison, W., R. J. Thomas, P. R. Krehbiel, T. Hamlin, and J. Harlin, 1999: A GPS-based three-dimensional lightning mapping system: Initial observations in central New Mexico. Geophys. Res. Lett., 26, 3573–3576, https://doi.org/10.1029/1999GL010856.
Rudlosky, S. D., and K. S. Virts, 2021: Dual Geostationary Lightning Mapper observations. Mon. Wea. Rev., 149, 979–998, https://doi.org/10.1175/MWR-D-20-0242.1.
Rudlosky, S. D., M. J. Peterson, and D. T. Kahn, 2017: GLD360 performance relative to TRMM LIS. J. Atmos. Oceanic Technol., 34, 1307–1322, https://doi.org/10.1175/JTECH-D-16-0243.1.
Rudlosky, S. D., S. J. Goodman, K. S. Virts, and E. C. Bruning, 2019: Initial Geostationary Lightning Mapper observations. Geophys. Res. Lett., 46, 1097–1104, https://doi.org/10.1029/2018GL081052.
Said, R. K., M. B. Cohen, and U. S. Inan, 2013: Highly intense lightning over the oceans: Estimated peak currents from global GLD360 observations. J. Geophys. Res. Atmos., 118, 6905–6915, https://doi.org/10.1002/jgrd.50508.
Siingh, D., P. R. Kumar, M. N. Kulkarni, R. P. Singh, and A. K. Singh, 2013: Lightning, convective rain and solar activity—Over the South/Southeast Asia. Atmos. Res., 120–121, 99–111, https://doi.org/10.1016/j.atmosres.2012.07.026.
Sun, H., J. Yang, Q. Zhang, L. Song, H. Gao, X. Jing, G. Lin, and K. Yang, 2021: Effects of day/night factor on the detection performance of FY4A Lightning Mapping Imager in Hainan, China. Remote Sens., 13, 2200, https://doi.org/10.3390/rs13112200.
Thiel, K. C., K. M. Calhoun, A. E. Reinhart, and D. R. MacGorman, 2020: GLM and ABI characteristics of severe and convective storms. J. Geophys. Res. Atmos., 125, e2020JD032858, https://doi.org/10.1029/2020JD032858.
Thompson, K. B., M. G. Bateman, and L. D. Carey, 2014: A comparison of two ground-based lightning detection networks against the satellite-based Lightning Imaging Sensor (LIS). J. Atmos. Oceanic Technol., 31, 2191–2205, https://doi.org/10.1175/JTECH-D-13-00186.1.
Virts, K. S., and W. J. Koshak, 2020: Mitigation of Geostationary Lightning Mapper geolocation errors. J. Atmos. Oceanic Technol., 37, 1725–1736, https://doi.org/10.1175/JTECH-D-19-0100.1.
Wang, C., Z. J. Luo, and X. Huang, 2011: Parallax correction in collocating CloudSat and Moderate Resolution Imaging Spectroradiometer (MODIS) observations: Method and application to convection study. J. Geophys. Res., 116, D17201, https://doi.org/10.1029/2011JD016097.
Williams, E. R., 2005: Lightning and climate: A review. Atmos. Res., 76, 272–287, https://doi.org/10.1016/j.atmosres.2004.11.014.
Williams, E. R., K. Rothkin, D. Stevenson, and D. Boccippio, 2000: Global lightning variations caused by changes in thunderstorm flash rate and by changes in the number of thunderstorms. J. Appl. Meteor., 39, 2223–2230, https://doi.org/10.1175/1520-0450(2001)040<2223:GLVCBC>2.0.CO;2.
Xian, D., P. Zhang, L. Gao, R. Sun, H. Zhang, and X. Jia, 2021: Fengyun meteorological satellite products for Earth system science applications. Adv. Atmos. Sci., 38, 1267–1284, https://doi.org/10.1007/s00376-021-0425-3.
Yang, J., Z. Zhang, C. Wei, F. Lu, and Q. Guo, 2017: Introducing the new generation of Chinese geostationary weather satellites, Fengyun-4. Bull. Amer. Meteor. Soc., 98, 1637–1658, https://doi.org/10.1175/BAMS-D-16-0065.1.
You, J., D. Zheng, Y. Zhang, W. Yao, and Q. Meng, 2019: Duration, spatial size and radiance of lightning flashes over the Asia-Pacific region based on TRMM/LIS observations. Atmos. Res., 223, 98–113, https://doi.org/10.1016/j.atmosres.2019.03.013.
Zhang, D., and K. L. Cummins, 2020: Time evolution of satellite-based optical properties in lightning flashes, and its impact on GLM flash detection. J. Geophys. Res. Atmos., 125, e2019JD032024, https://doi.org/10.1029/2019JD032024.
Zhang, D., K. L. Cummins, P. Bitzer, and W. J. Koshak, 2019: Evaluation of the performance characteristics of the Lightning Imaging Sensor. J. Atmos. Oceanic Technol., 36, 1015–1031, https://doi.org/10.1175/JTECH-D-18-0173.1.
Zhang, D., T. Lang, K. Cummins, and M. Quick, 2021: Inter-comparison of ISS-LIS, TRMM-LIS, and GLM. GLM Science Meeting, Huntsville, AL, NASA.
Zhang, W., W. Hui, W. Lyu, D. Cao, P. Li, D. Zheng, X. Fang, and Y. Zhang, 2020: FY-4A LMI observed lightning activity in Super Typhoon Mangkhut (2018) in comparison with WWLLN data. J. Meteor. Res., 34, 336–352, https://doi.org/10.1007/s13351-020-9500-4.
Zhang, Y., and Coauthors, 2022: Advances in lightning monitoring and location technology research in China. Remote Sens., 14, 1293, https://doi.org/10.3390/rs14051293.
Zheng, D., and D. R. MacGorman, 2016: Characteristics of flash initiations in a supercell cluster with tornadoes. Atmos. Res., 167, 249–264, https://doi.org/10.1016/j.atmosres.2015.08.015.
Zheng, D., Y. Zhang, Q. Meng, L. Chen, and J. Dan, 2016: Climatological comparison of small- and large-current cloud-to-ground lightning flashes over southern China. J. Climate, 29, 2831–2848, https://doi.org/10.1175/JCLI-D-15-0386.1.
Zhu, Y., V. A. Rakov, M. D. Tran, and A. Nag, 2016: A study of National Lightning Detection Network responses to natural lightning based on ground truth data acquired at LOG with emphasis on cloud discharge activity. J. Geophys. Res. Atmos., 121, 14 651–14 660, https://doi.org/10.1002/2016JD025574.
