Lightning strike at night

ISS-LIS data analysis based on LMA networks in Europe

Lightning strike at night
Lightning strike at night

This study supports MTG-LI Cal/Val preparations, and focuses on comparisons between measurements made with the European LMA (Lightning Mapping Arrays) and ISS-LIS (the Lightning Imaging Sensor flying on the International Space Station).

Last Updated

14, November 2020

Lightning is a sudden electrostatic discharge that occurs typically during a thunderstorm. This discharge occurs between electrically charged regions of a cloud (intra-cloud lightning or IC), between two clouds (CC lightning), or between a cloud and the ground (CG lightning).

The three categories of lightning here mentioned define the complete lightning activity.

Lightning is a source of different type of signals in the electromagnetic spectrum, in detail: Very High Frequency signals (VHF), Very Low Frequency signals (VLF), Low Frequency signals (LF), and finally optical pulses. These signals can be detected by different type of instruments and can be employed for the location in space and time of lightning, as well as for their physical characterisation. Different instruments are capable of detecting either a part of or all the lightning categories listed above.

Only a clear understanding of the specificity of the retrieval provided by the different detection techniques/instruments can enable the use of measurements of different electromagnetic signals for a harmonised and complementary description of lightning phenomena.


The main objective of this study is to perform a detailed cross-analysis of the data acquired by:

  • ISS-LIS: the Lightning Imaging Sensor, hosted by the International Space Station since 19 February 2017 (LIS was on board the Tropical Rainfall Measuring Mission (TRMM) from 1998 to 2015), is an instrument which performs observations of the total lightning activity (i.e. CC + IC + CG) in the optical band. The detection principle employed by this instrument defines the one of the Meteosat Third Generation Lightning Imager (MTG LI).
  • LMA networks: the Lightning Mapping Arrays detect VHF signals from lightning, and allow for a very detailed 3D mapping of these phenomena. LMA have small Field of View (FOV), with typical size of tens of kilometres, in which they detect the total lightning activity.

The study will answer the following questions:

  1. How can LMA data be employed to validate and, if needed, correct/complement the information on lightning events, groups, and flashes provided by ISS-LIS?
  2. How can LMA data and ISS-LIS data be used to define a new and refined statistical description of lightning pulses, groups, and flashes?
  3. What are the performances of ISS-LIS that are evaluated against LMA measurements?
  4. How can LMA data and ISS-LIS data be used to define test cases for the EUMETSAT MTG LI end-to-end processor?
  5. From the lesson learned by comparing ISS-LIS data against LMA data: how can LMA data be employed in the validation of MTG LI during commissioning and in routine monitoring?


In order to answer to the questions listed in the objectives the following tasks were undertaken:

Task 1: ISS – LIS v LMA

In this task the two instruments have been compared in many details related to both technology and design that are crucial in defining the observation properties. Both differences between/ synergies of the instruments were highlighted. Section 1 of the Final Report provides the reader with many details on the two instruments.

Task 2: Data for the analysis

In this task two 'metrics' for selecting observations to be used for the ISS-LIS v LMA comparison were defined. Both are based on descriptors of LMA performance: the first one uses the maximum number of sources per LMA flash, and the second one uses the LMA sensitivity to power levels of sources. Both performances degrade with the distance from the centre of the LMA network. The data selection is performed by combining these descriptors with the number of ISS-LIS flashes for different LMA performance regions: the larger the area around the LMA network centre, the higher the number of ISS-LIS flashes that can be used for the analysis. In the data selection, it all comes down to a compromise between good LMA performances and high-enough number of ISS-LIS flashes. The selected datasets for different distances from the centre of the LMA network are provided in Section 2.5 of the Final Report.

Task 3: Processing of data

In this task the processing of the data was undertaken. The key results worth highlighting here are provided in the bullet points below. Section 4 of the Final Report provides the reader with all the details of the processing outcome.

  1. ISS-LIS flash detection efficiency against LMA
    The average flash detection efficiency of ISS-LIS is about 70%. Moreover, the detection efficiency rapidly drops when the reference LMA sources in the flashes do not reach high cloud altitudes (around 10 km). The decay is relevant, from 70% to about 20%. This is due to optical signal attenuation through the cloud. In Figure 1 two examples of the flash detections are provided.
    Representation of ISS-LIS detections
    Figure 1: Representation of ISS-LIS detections (events, top panels with black dots) and LMA detections (sources, bottom panels with red dots) for two different flashes on 20180917 and 20180918, respectively left and right columns.
  2. Flash duration
    The flash duration is defined as the time difference between the last and first event/detection. This definition applies to both ISS-LIS and LMA flashes. When comparing the average properties of the selected flashes one finds out that the LMA detect, on average, longer flashes than ISS-LIS. In detail, ISS-LIS flashes are about 20%–30% shorter than LMA flashes. Moreover, ISS-LIS systematically detects the beginning of the flash later than LMA (always within the first 20% of the total duration of the LMA flash) and does not manage to capture the final part of the flash (see Figure 2).
    Representation of the detection of same flash from LMA
    Figure 2: Representation of the detection of same flash from LMA (coloured dots) and ISS-LIS (black open cicles); on the x-axis the time in seconds, on the y-axis the height in km. The figure provides an example of the typical difference between the flash duration and relative detection from the two instruments.
  3. Location accuracy
    The location accuracy is derived from: i) the off-set between LMA and ISS-LIS detections, and ii) the difference in the overlap of the detections (see for examples Figure 1). An average offset of about 5 km is found between the flash centres of the two instruments. This is well below the size of the ISS-LIS pixel (of about 8 km), indicating that the location accuracy of ISS-LIS is very good. When comparing the overlap of the detections the variability of the results is large. On average, 47% of the LMA flashes is covered by the ISS-LIS detections. Also, one finds that a similar percentage of ISS-LIS detections are not covered by LMAs. The result from the overlap analysis could be explained by considering the different nature of the signals detected by LMA and ISS-LIS respectively (see description in the Objectives section).

Task 4: Preparing for MTG LI

In this task two activities were undertaken:

  1. Derive flash properties to be used to define inputs for the EUMETSAT end-to-end processor, in the context of the pre-flight Level 2 performance assessment. The following properties were provided:
    • flash duration
    • flash size
    • inter-flash time
    • relation between flash duration and size
  2. Outline the methodology for the assessment of the in-flight performances of LI, based on the work done on the comparison ISS-LIS v LMA. The assessment of the following performance was discussed, and for each of them the confidence of success was provided. From this one learns that LMA data will be important for the assessment of the LI:
    • flash detection efficiency
    • timing accuracy
    • location accuracy
    Moreover, a strategy for the evaluation of the LI performances with the aid of LMA data are outlined. LMA data can be employed for the assessment of LI performances in different phases:
    • in early observation phase, for the identification of reference cases which may serve for detection and processing tuning purposes, instrument validation, and performance assessment;
    • during commissioning, to follow up the activity on the tuning and to perform an extensive validation;
    • during routine operations, for instrument monitoring, product validation.
    For the different uses, the LMA data would be provided on a daily basis, on a weekly basis, and on a monthly basis, respectively.