The Multiplatform Precipitation Feature (MPF) Database: A Storm‐Centric Synthesis of Space‐ and Ground‐Based Precipitation and Lightning Data Sets for Convective Studies

This paper outlines the construction of a new Multiplatform Precipitation Feature (MPF) framework designed to facilitate storm‐based analyses of data sets from multiple independent observing platforms for convective studies. The MPF approach is applied to ground‐ and space‐based precipitation observations and retrievals from the Global Precipitation Measurement mission Validation Network with space‐based lightning measurements from the Lightning Imaging Sensor on board the International Space Station to create a prototype MPF database. The data are synthesized in a thunderstorm‐like, feature‐based framework that enables broad statistical analyses of the storm‐based relationships between the microphysics, kinematics, and electrical properties of convection that can contribute insight into their morphological, regional, and seasonal variations. This manuscript details how MPFs are defined from precipitation data and how observations from multiple additional platforms are combined within the context of an MPF thunderstorm boundary and used to characterize the feature. A demonstration of the prototype database as a proof‐of‐concept showcases its flexibility to conduct both large‐scale storm‐based statistical studies as well as MPF‐facilitated thunderstorm case studies.

A rich history of detailed thunderstorm case studies has demonstrated and highlighted foundational links between the electrical processes, microphysics, and dynamics of deep convection (e.g., Bruning & MacGorman, 2013;Bruning et al., 2010;Calhoun et al., 2013;Carey & Rutledge, 1998;Carey et al., 2019;Deierling & Petersen, 2008;Deierling et al., 2008;Dye et al., 1974Dye et al., , 1986Dye et al., , 1989;;Fuchs & Rutledge, 2018;Fuchs et al., 2018;Lang & Rutledge, 2002;Lhermitte & Williams, 1985;Reap & MacGorman, 1989;Wiens et al., 2005).These studies have employed ground-based instrumentation such as Lightning Mapping Arrays (LMAs, Rison et al., 1999) and single-and dual-polarization Doppler radars that offer fine spatial and temporal resolution for process studies at and below the spatial scale of convective updrafts.However, they are restricted in observational range and in the case of four-dimensional lightning detection, limited to a small number of regional networks.These localized samples limit the broader insights into the nature of deep convection and extrapolation of the regional and seasonal variability in complex relationships.et al., 2002;Williams & Stanfill, 2002).Such large global data sets are necessary to perform broad statistical analyses to assess the relationships between the environment, the developing convection, and the potential for lightning.The broad perspective, by necessity, leaves out the detailed storm-level convective context critical to understanding a storm's electrical properties because of the inextricable link between lightning and the convective updraft.
The construction of a new method of defining thunderstorm-centric features embodies a fusion of the broader spaceborne approaches and detection capabilities with process-based, storm-level analysis practices.The Multiplatform Precipitation Feature (MPF) framework described herein combines global lightning and precipitation information, facilitating statistical assessment of storm-based relationships between lightning and deep convective microphysics.This approach is unique in that it maintains the storm-level context among disjoint global, gridded data sets.It is the storm-level context that enables lightning to be used successfully as a convective diagnostic and climate variable.Refining the relationships between lightning energetics and convective parameters enables the retrieval of cloud and precipitation properties from quantitative lightning data.

The Lightning and Convection Connection
The convective updraft is the lifeblood of a thunderstorm: its development and interaction with the surrounding environment determine the characteristics of the storm, its cloud and precipitation microphysics, and properties of lightning flashes such as their size, frequency, and location with respect to the updraft.Clouds primarily become electrified via the noninductive ice-ice collision mechanism (Takahashi, 1978), which by definition requires the presence of small ice and actively riming graupel particles colliding in a turbulent updraft.Observational research has supported lab-based electrification hypotheses through analysis of the co-evolving microphysical, kinematic, and electrical relationships in deep convection.A storm's ability to develop lightning closely follows the development of a sufficiently strong convective updraft, characterized by vertical velocity measurements of at least 7 m s −1 (Dye et al., 1986;Zipser & Lutz, 1994).The updraft then supplies the supercooled liquid water needed for graupel growth and electrification (Dye et al., 1986;Krehbiel et al., 1979) and supports gravitational separation of charged ice particles toward cloud-scale charge structures and the development of sufficient electrical fields for lightning discharge (Dye et al., 1974(Dye et al., , 1989;;Lhermitte & Williams, 1985).
Characteristics such as updraft width, depth, and speed link closely with increases in charged hydrometeor populations, which affect the spatial structure and strength of electric fields and the lightning flashes they produce (Bruning & MacGorman, 2013;Carey & Rutledge, 1996;Carey et al., 2019;Deierling & Petersen, 2008;Deierling et al., 2008;MacGorman et al., 1981;Stolzenburg et al., 1998;Williams, 1985).Specifically, increases in the updraft size and augmentation of ice precipitation processes increase the number of lightning flashes and shift their size distribution toward smaller flashes within convective regions of a thunderstorm (e.g., Brothers et al., 2018;Bruning & MacGorman, 2013;Stolzenburg et al., 1998).Additional relationships between particle advection, the establishment of layered charge, and coincident flash behavior have also been gleaned from analyses of non-convective regions, such as the trailing stratiform region of mesoscale convective systems (Stolzenburg et al., 1998).These relationships have been leveraged for lightning safety applications, including identification of lightning flashes more likely to connect with the ground (e.g., Schultz et al., 2017Schultz et al., , 2021)).They also support lightning as a useful metric of updraft intensification that precedes the onset of severe phenomena such as hail, strong winds, and tornadoes (Gatlin & Goodman, 2010;Schultz et al., 2009;Williams et al., 1999).
Although much of the aforementioned literature focuses on the relationship between lightning and meteorological conditions for the purposes of prediction, lightning itself is also a climatological variable (Koshak et al., 2014(Koshak et al., , 2015)), and a possible indicator for long-term climate change (Lavigne et al., 2019;Seeley & Romps, 2015;Williams, 2020).Lightning alone typically serves as a diagnostic of a characteristically strong updraft and the existence of graupel and supercooled water, as discussed above (Deierling et al., 2008;Takahashi, 1978;Yoshida et al., 2017).Deeper analyses of interrelationships between the electrical characteristics of lightning, cloud and precipitation microphysics, and convective kinematics are critical to advancing the retrieval of cloud and precipitation characteristics from lightning data and the further use of lightning as a geophysical variable in its own right.

The Role of Multiplatform Precipitation Features
The Multiplatform Precipitation Feature (MPF) framework described herein was constructed to capture these interrelationships through the synthesis of relevant data into storm-based features.Specifically, each MPF combines satellite-and ground-based radar observations and retrievals to address microphysics and kinematics (e.g., ice content and updraft size and speed) along with space-based lightning measurements (e.g., flash rate, flash size, radiance, etc.) in a single storm boundary.MPFs effectively aggregate numerous well-observed thunderstorm cases, enabling statistical storm-based analyses of convective microphysics, kinematics, and electrical properties on a scale far exceeding that of case studies in the literature.
The data sets involved in creating MPFs are outlined in Section 2. Section 3 details the construction of an MPF, the characteristics that are added to MPFs from each contributing data set, as well as the process by which independent data sets are synthesized into the MPF framework.We then demonstrate the flexibility of the MPF approach to enable statistical analyses as well as an example MPF-facilitated thunderstorm case study in Section 4. Finally, we discuss the future of the MPF framework, expansion to other precipitation and lightning data sets, and the science it enables in Section 5.

Data
The prototype MPF database we present here consists of storm-based features identified using data from the Dual-frequency Precipitation Radar (DPR) on board the NASA Global Precipitation Measurement (GPM) Core Observatory satellite (Hou et al., 2014).The thunderstorm characteristic data that populate each MPF in the database are collected from the DPR, ground-based weather radar data captured by the GPM Validation Network (VN, Gatlin et al., 2020), and lightning measurements from the Lightning Imaging Sensor on board the International Space Station (ISS LIS, Blakeslee et al., 2020).An example of a GPM DPR/ISS LIS coincidence over ground-based radars in the VN with dual-Doppler analysis capability is shown in Figure 1 for thunderstorms over western Louisiana on 25 May 2020.

The GPM Validation Network
The international GPM mission was launched to collect global space-based measurements of rain and snow from an intercalibrated network of satellites (Hou et al., 2014).Measurements from instruments on board the central GPM Core Observatory satellite, including the DPR, serve as the reference measurement standard among the GPM mission constellation.The GPM Core Observatory flies in low Earth orbit (LEO) at an inclination of ±65°, measuring most of the globe's precipitation.Satellite data and algorithm validation efforts have been conducted as part of the mission since its inception in 2014 (Petersen et al., 2020), undertaken using an extensive Ground Validation System and VN of operational and research dual-polarization Doppler radar data and analysis software (GSFC, 2015).
The VN software geometrically matches concurrent space-and ground-based precipitation observations made during GPM satellite overpasses of ground-based weather radar sites (Schwaller & Morris, 2011).The VN is primarily used for robust validation of warm rain products and retrievals for the GPM mission.Therefore, GPM measurements of frozen precipitation are filtered before VN processing.Coincident ground-based dual-polarization radar measurements during rainy GPM overpasses are quality controlled through a combination of automated and manual techniques (Pippitt et al., 2015).These data are used to obtain a statistical description of the precipitation column within the satellite's instantaneous field of view.Geophysical variables retrieved from the ground-based radars include rainfall rate, the raindrop size distribution (Tokay et al., 2020), and hydrometeor types (Dolan et al., 2013;Dolan & Rutledge, 2009).Recently, three-dimensional winds have been added to this list for several ground-based radar pairs conducive to multi-Doppler retrieval, which is done using the three-dimensional variational technique implemented in the Pythonic Direct Data Assimilation (PyDDA, Jackson et al., 2020;Shapiro et al., 2009;Potvin et al., 2012) software.The resulting synthesized data product, referred to hereafter as a "VN matchup," pairs satellite and ground radar measurements and incorporates ground-based radar retrievals (e.g., dual-polarization data-based hydrometeor type identification and Doppler-derived three-dimensional winds) that otherwise unattainable from satellite instruments.VN matchup data and software are archived and publicly available through the GPM mission Ground Validation web portal.The current study makes use of GPM VN data from March 2014 through August 2022.

The Lightning Imaging Sensor on Board the International Space Station
The LIS instrument measures lightning at cloud-top from spaceborne platforms during both day and night by time-differencing geolocated images at the 777.4 nm wavelength oxygen absorption line (Blakeslee et al., 2020).LIS was first flown on board NASA's Tropical Rainfall Measuring Mission (TRMM, Kummerow et al., 1998) satellite.In 2017, a flight spare LIS was adapted and employed on the International Space Station (ISS) as part of the fifth Space Test Program -Houston (STP-H5) sponsored by the United States Department of Defense.The ISS is currently in LEO at ≈400 km elevation with an inclination of ±52°(LIS extends to ±55°) (Blakeslee et al., 2020).LIS lightning observations have ≈4 km spatial resolution and 2 ms temporal resolution.
Quality-Controlled Version 2 ISS LIS lightning data (Lang, 2022) are publicly available from the NASA Global Hydrometeorology Resource Center (GHRC).The lightning data are broken into four clustering levels: events (individual 4 km pixels exceeding the background value), groups (clusters of events within a 2 ms frame), flashes (clusters of groups close in time and space), and areas (contiguous regions of lightning in a single orbit) (Mach et al., 2007), each described with geographical, temporal, and energetic characteristics (i.e., radiance, footprint, and energy density).

Defining an MPF
The format of MPFs and their construction follows from the Precipitation Feature Database, which enables large-scale analysis of large spaceborne precipitation data sets (Liu et al., 2008;Nesbitt et al., 2000).Unlike the Precipitation Feature Database, MPFs focus on convective thunderstorm cells.Figure 2 traces the process of individual MPF construction on a VN matchup event that was coincidentally observed by the ISS LIS.Data from 2017 through 2022, encompassing availability of data from both the VN and ISS LIS, are included in this proof of concept framework construction.
MPFs are identified using equivalent GPM DPR radar reflectivity factor data captured within a VN matchup.Reflectivity data from the GPM DPR are used because they are consistent and available globally, enabling future extension of the MPF framework outside of the VN and into oceanic regions where ground radar data are not available.First, reflectivity data are interpolated to the −10°C environmental temperature height (determined from local observations or model reanalysis, also included in the VN) (Figure 2a).The −10°C height is the level at which noninductive charging is most active and gives the greatest insight into the microphysics associated with electrification and lightning processes (Carey & Rutledge, 1998).Using a −10°C threshold to identify MPFs also excludes shallow precipitation that is not relevant for studying deep and potentially electrified convection.An image segmentation and labeling algorithm from the SciPy multidimensional image processing package (Virtanen et al., 2020) is then applied to the DPR reflectivity data at −10°C to identify distinct reflectivity regions (Figure 2b).Smaller regions are typically interpreted as individual convective cores while larger, more conglomerate regions may represent adjacent groups of convective cores between which ice microphysical and charging processes are shared, contributing to more complex electrification structures.
The segmented reflectivity regions must meet several criteria to be considered precipitation features.First, regions must consist of at least four contiguous DPR reflectivity pixel with precipitation at −10°C (following Liu  , 2012).For reference, each VN matchup voxel (i.e., three-dimensional pixel) size is variable, with an approximate ground-projected area of 25 km 2 .Therefore, a minimum feature region size of ≈100 km 2 roughly relates to the horizontal size scale of convection.It should be noted that VN geomatch voxel sizes vary because the combined, resampled data structure depends on variable intersections between the satellite viewing angle and ground radar scans.This variability ensures the closest data pairing without any interpolation, though it sacrifices a regular gridded data structure (Schwaller & Morris, 2011).Segmented reflectivity regions meeting the minimum size of four DPR pixels are next prescribed a boundary based on the region's convex hull (Devadoss & O'Rourke, 2011) of outermost connecting points (Figure 2b).At this stage, any smaller features that are wholly contained within the convex hull of another are discarded.Each remaining reflectivity region is hence defined as an MPF and its convex hull serves as the spatial boundary within which properties from additional data sets are extracted.

Incorporating Radar Data and Retrievals Into MPFs
Data from VN matchups are populated to MPFs in the form of summary statistics, including GPM DPR Ku-band radar reflectivity and ground-based radar retrievals from each VN matchup.Within an MPF, the minimum and maximum DPR reflectivity values and the average height at which they occurred; the mean, minimum, and maximum reflectivity values at the interpolated heights of −10°C and −20°C; and the maximum heights at which 20, 30, 40, and 50 dBz reflectivity occurred are stored.
Because the resolution of DPR footprints is coarser than the ground-based radar data they are matched with, ground radar properties are stored in VN data as representative statistics.That is, each DPR footprint pairs with a mean, minimum, maximum, and standard deviation value of the corresponding ground radar data.Therefore, several properties retained in MPFs are provided as various representations of these statistical fields.For instance, because VN geomatch data contain fields such as mean, minimum, and maximum vertical motion, MPFs store the minimum, mean, and maximum of each of these that are found within the feature boundary and the average heights at which they occur (e.g., mean of the maximum vertical velocity field in the feature, minimum of the minimum vertical velocity field in the feature, etc.).The size of the vertical motion field where values exceed 5 and 10 m s −1 , referred to as the 5 m s −1 updraft volume and 10 m s −1 updraft volume, respectively, are also added to the MPF for each of the minimum, mean, and maximum vertical velocity fields from the VN geomatch data.Profiles of these statistical fields are extracted from the VN matchup and added as well.Similarly, though a single most likely hydrometeor type is identified for each ground radar data point, referred to as a radar gate, a group of radar gates with multiple different identified hydrometeor types is associated with each DPR footprint.Therefore, the VN geomatch data report the total number of radar gates as well as the number of gates identified as each of 10 hydrometeor types within each DPR footprint.This information is stored to the MPF in the form of vertical profiles and includes the number of radar gates within the MPF, the fraction of gates associated with non-precipitation-sized ice hydrometeor types (i.e., ice crystals and vertically oriented ice categories), the fraction of gates associated with precipitation-sized ice hydrometeor types (i.e., graupel and hail hydrometeor categories), and the heights at which they occur.
A complete list of the radar-based information saved to each MPF is provided in Tables S1 through S3 in Supporting Information S1.

Incorporating Lightning Data Into MPFs
ISS LIS data (Lang, 2022) are added to each MPF by finding any coincident LIS overpasses (with or without lightning) within ±5 min of the radar timestamp within the MPF boundary using polygon matching from the Motley software suite (Hadfield, 2001).The ±5-min window encompasses the volume sampling time of ground-based radars in the VN (see Figure 2c).If any lightning occurred, the geographic and temporal information, and electrical characteristics for each event, group, flash, and area within the MPF boundary are stored to the MPF (Figure 2d).The ISS LIS variables saved to each MPF are listed in Table S4 in Supporting Information S1.

Demonstration of the Prototype MPF Database
The construction of the individual MPFs, the attributes stored to them, and the collective database enable a range of analyses.The composite database facilitates broad, statistical analysis of thunderstorms, while individual storm events can be reconstructed using the information associated with each MPF for traditional case-study level analysis.Though not the focus of this proof-of-concept study, a few illustrative examples demonstrate the capabilities of this kind of data set.

Broad Statistical Enabled Analysis
The MPF framework of combined radar and lightning image data sets facilitates the analysis of broad statistical variables such as feature minimum, maximum, or mean characteristics of any stored thunderstorm property.For example, we compare the maximum GPM DPR Ku-band DPR reflectivity at −10°C for MPFs with and without lightning observed in the coincident LIS field of view in Figure 3.These distributions support the idea that electrified convection is more typically associated with greater reflectivity values in the mixed-phase region, consistent with the presence of riming graupel (see Deierling et al., 2008;Williams et al., 1989;Yoshida et al., 2017).
It is also possible to examine the LIS-observed electrical characteristics in the context of radar-derived feature properties.In Figure 4, we show the lightning event footprint sizes for MPFs with LIS-observed lightning in three categories of maximum reflectivity at −10°C: 15-30 dBz, 30-45 dBz, and 45-60 dBz.Note the shift to much larger footprint sizes for MPFs with lower maximum reflectivity (i.e., <30 dBz at −10°C).While further analysis of these data is beyond the scope of this proof-of-concept demonstration of the data set, we speculate that this signal of larger lightning footprints in areas of lower reflectivity may correspond with lightning in stratiform precipitation.The order of magnitude fewer flashes in lower maximum reflectivity regions, compared to the distributions for higher reflectivities, paired with the shift to larger flash sizes, is consistent with case-study findings in the literature that lightning flash sizes increase the further they are displaced from the convective core, and lighting flashes are fewer and larger in stratiform and anvil precipitation relative to the convective region (Bruning & MacGorman, 2013;Carey et al., 2005).

Case Study Enabled Analysis
The MPF metadata offer traceability to the original VN matchup, its DPR and ground radar data components, and coincident ISS LIS data.Using these source data, we present an example case-study level analysis in Figure 5 of ISS LIS individual lightning data in the context of ground-based radar reflectivity, retrieved vertical wind, and 10.1029/2023EA003137 8 of 12 classified hydrometeor types.Quick comparisons of the radar reflectivity structure and the locations of ISS LIS events show that the southeastern portion of the feature was actively producing lightning at the time of the ISS LIS overpass (Figure 5a), while the vertical motion and identified hydrometeor types confirm that most lightning was associated with a mature updraft and established graupel regions (Figures 5b and 5c).Of interest for further analysis is that the southeastern portion of the updraft was associated with more active lightning than the stronger updraft region to the northwest within the feature.The ability to retrieve the data used to create the feature from its metadata provides information for a user to pursue additional temporal analysis of storms of interest as desired.

Conclusions and Future Work
This manuscript details the creation of a Multiplatform Precipitation Feature (MPF) framework applied to coincident Global Precipitation Measurement (GPM) spaceborne radar, ground-based dual-polarization Doppler radar data from the GPM Validation Network (VN), and Lightning Imaging Sensor data from on board the International Space Station (ISS LIS).We construct a feature boundary, or MPF, based on contiguous areas of precipitation sampled by the GPM Dual-frequency Precipitation Radar (DPR) at −10°C.Each MPF stores summary statistical data within the boundary of precipitation microphysics and, where possible, retrieved vertical winds from radar data along with lightning observations from the ISS LIS (Figure 2).We provide detailed lists of data saved within each MPF in Supporting Information S1.
Synthesizing data from multiple instruments into an MPF enables the simultaneous analysis of thunderstorms' microphysical, kinematic, and lightning energetic properties.This then creates an opportunity to assess broad statistical relationships between these intertwined thunderstorm characteristics while maintaining the context and structure of individual thunderstorms.This approach overcomes contrasting limitations of traditional analysis approaches.Historical thunderstorm case studies employ instruments offering excellent spatial and temporal resolution but are sparse and highly localized, limiting the ability of their observations to inform statistical relationships absent regional biases.Conversely, satellite-based lightning observations used to establish broader lightning-convection relationships offer global coverage but lack spatial and temporal detail that informs about deep convective processes.The MPF structure maintains the storm-level context that is crucial for deconstructing the physical processes occurring within it but without the limited range of fine-scale sampling.As lightning is closely tied to the properties of the convective updraft and its microphysics, it is an excellent diagnostic tool for the lifecycle and intensity of a convective storm (e.g., Goodman & MacGorman, 1986;Williams et al., 2005).It is also recognized as a geophysical variable and designated as an essential climate variable by the World Meteorological Organization (Global Climate Observing System, 2016).The framework of MPF allows for the analysis of the relationships between the lightning energetics, storm kinematics, and microphysics from the storm-centric perspective of ground-based analysis, but with the wider coverage and statistical capabilities of spaceborne data sets.

Ongoing and Future Work
We outline the development of a method to construct MPFs that synthesizes data from multiple moving platforms in space with observations and retrievals from static ground-based instruments in a thunderstorm-centric framework.The first iteration of the MPF database pairs GPM DPR Ku-band observations with scaled, coincident ground-based radar data sets from the GPM VN.These publicly accessible, quality-controlled data include post-processed retrievals, making them readily available, low-cost candidates for this type of proof-of-concept study.However, the requirements of (a) the existence of precipitation within a GPM overpass that (b) includes multiple radars in the VN suitable for three-dimensional wind retrieval along with (c) a coincident ISS LIS overpass returns a small sample of MPFs with the complete suite of lightning, precipitation microphysics, and kinematics, as shown in Figure 3.
Following the first successful proof-of-concept of platform synthesis presented herein, there is ongoing work to define MPFs using ground-based radar alone, without the restriction of a GPM overpass.Expanding beyond the GPM VN data set requires additional pre-processing to quality control the radar data and retrieve algorithmically derived products such as three-dimensional winds and hydrometeor identification as is accomplished within the VN.Defining MPFs from ground-based radar implies that this version of features cannot be obtained over the ocean, as with DPR-defined features.However, the change offers several benefits for convective studies.First, solely by eliminating the required coincidence of the GPM, preliminary results yield thousands of MPFs, expanding upon the size of the existing proof-of-concept VN-based MPF database by two orders of magnitude.Additionally, ground-based radar data and retrievals included in each non-VN MPF are not scaled to the DPR footprint resolution as occurs in the VN matchup process, offering more detail and more specific quantification of precipitation microphysics and kinematics.Expanding the MPF database beyond the GPM VN also affords the ability to increase the sample of winter convection.This larger, more complete database is expected to enable more robust statistical analyses of storm-based relationships between the microphysical, kinematic, and electrical properties of convection.The MPF development method we outline in this manuscript establishes a foundational infrastructure to synthesize multiple spaceborne and ground-based precipitation and lightning data sets into a storm-centric framework.The MPF framework can be adapted and scaled up to other satellite-and ground-based observational data sets, as with the ground-based radar described above.Anticipated future work includes development of quantitative retrievals of thunderstorm characteristics based solely on lightning data for diagnosis of deep convection in the absence of other satellite and radar data.

Figure 1 .
Figure 1.Global Precipitation Measurement (GPM) Dual-frequency Precipitation Radar (DPR) (gray), ISS LIS (turquoise), and VN dual-Doppler domain for the KLCH and KPOE radars (pink) coincidence over a storm in western Louisiana on 25 May 2020.The GPM DPR Ku-band reflectivity is overlaid, and LIS-detected lightning flashes shown in white.

Figure 2 .
Figure 2. Flow chart describing how a Multiplatform Precipitation Feature (MPF) is constructed.(a) Global Precipitation Measurement Dual-frequency Precipitation Radar Ku-band reflectivity data are gridded to height of −10°C and filtered to segment convective cores.(b) The Scikit-learn ndimage algorithm is applied to identify and label distinct reflectivity regions.Convex hulls are determined around ndimage regions, defining the feature boundary.(c) Coincident ISS LIS overpasses are identified within the boundary and ±5 min from the radar timestamp.(d) If lightning is detected, the event, group, flash, and area data from within the boundary are appended to the MPF.

Figure 3 .
Figure 3. Frequency distribution of feature-maximum reflectivity at −10°C derived from the Global Precipitation Measurement Dual-frequency Precipitation Radar for all VN Multiplatform Precipitation Features (MPFs).The distribution of MPFs that were observed by the Lightning Imaging Sensor on board the International Space Station (ISS LIS) but did not have lightning is plotted in black, and that for MPFs with ISS LIS-observed lightning are shown in red.

Figure 4 .
Figure 4. Frequency distribution of lightning event footprint size for all Lightning Imaging Sensor on board the International Space Station lightning events found within an Multiplatform Precipitation Feature (MPF) boundary.The distributions are broken into three ranges of MPF maximum reflectivity at −10°derived from the Global Precipitation Measurement Dual-frequency Precipitation Radar: 15-30 dBz (blue), 30-45 dBz (red), and 45-60 dBz (black).

Figure 5 .
Figure 5. Case-study level overview of the southernmost Multiplatform Precipitation Feature (MPF) identified on 25 May 2020 in western Louisiana.The MPF was defined based on Dual-frequency Precipitation Radar reflectivity data but encapsulate ground-based radar data and retrievals that provide increased detail.Lightning Imaging Sensor on board the International Space Station lightning event locations are shown overlaid on KLCH S-band (a) gridded radar reflectivity at 5 km, (b) vertical velocity, and (c) dual-polarization hydrometeor identification.