the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 May 2024
The World Wide Lightning Location Network (WWLLN) over Spain
Abstract. A study to determine the detection efficiency and location accuracy of the Worldwide Lightning Location Network (WWLLN) over Spain is presented by comparing data with those of the Meteorological Spanish Agency (AEMET), taken as a ground truth. The WWLLN operates a planetary distributed network of stations which detect lightning signals at a planetary scale. Very high currents from lightning strokes radiate strong Very Low Frequency (VLF) signals in the band 6–22 kHz, which are detected up to 10,000 km by the WWLLN stations. Two WWLLN stations operate in the Iberian Peninsula since 2012, which are supported by other stations at distances below 4000 km. The stations in the Iberian Peninsula are at a distance of around 800 km. This is a short distance in comparison with the typical distance between WWLLN stations in other areas, which is around 5.000–15.000 km. The WWLLN stations locate the time and position of the lightning stroke detected by triangulation, similarly as Global Positioning Systems do. Distances to each station are obtained by means of the time of arrival of the signal to the corresponding stations. A lightning detection is considered as a valid one when at least five stations detect it with a time and space coincidence with AEMET data of 0.5 s and 20 km, respectively. A study of the WWLLN performance for the whole area of Spain is carried out, obtaining that the detection efficiency of WWLLN is around 38 % with a location accuracy between 2 and 3 km. The efficiency for high energy strokes is considerable higher. The results obtained for Spain are better than those obtained in previous studies in other areas of the World, which may be caused by the high density of stations in the Spanish region and its surroundings. A study for two reduced regions with different geographic features is also considered to assess the possible influence of the different typology of storms on the network features. Finally, an application of the WWLLN data for three major storms in 2020, 2021 and 2022 in the Mediterranean area of Spain demonstrates that the WWLLN is well suited for tracking the time evolution of adverse meteorological phenomena.
- Preprint
(1779 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 28 Jun 2024)
-
RC1: 'Comment on egusphere-2024-704', Anonymous Referee #1, 11 Jun 2024
reply
The authors compare overlapping lightning detection networks, specifically the regional lightning location system (LLS) operated by AEMET and the global LLS WWLLN, in terms of detection efficiency and location accuracy, using data from a four-month period. However, this manuscript falls short in several areas, as outlined below.
Concerns:
- The data spans only a four-month period (January 1 – April 30, 2012) and is outside the typical thunderstorm season, encompassing a total of only 20,651 lightning strokes within the yellow polygon (and drops further to 3389 strokes in the green square, and an even lower amount of detections within the cyan rectangle which is not being mentioned). The rationale provided is that the AEMET dataset for that time was available as open data. Given that it is now 2024, a more recent dataset would be significantly more valuable. This is not only because it would be more current but also because the AEMET network from 2012 differs from the present-day network. The justification for using this specific dataset, as mentioned in lines 337-339, is insufficient. The authors acknowledge the WWLLN network for providing the lightning location data used in this paper. I don’t see why the authors cannot ask a more recent data set to AEMET. Moreover, the time resolution within this open dataset is only one second, making it unusable for a comparison study.
- The manuscript is submitted to NHESS, which is dedicated to research on natural hazards and their consequences. Although the current manuscript deals with lightning observations, it is essentially a comparison study. Therefore, the more appropriate journal for such a type of study would be EGU Atmospheric Measurement Techniques, which includes intercomparison research.
Specific remarks:
- In the abstract and the text, it should be clarified whether the DE pertains to strokes or flashes.
- In several parts of the text, it is stated that the primary objective of regional and national lightning location networks is to detect CG strokes, with IC events being of lesser importance. I disagree with this. For example, air traffic controllers are very interested in IC activity and usually receive the observations through a LLS network provider.
- The introduction mentions another global network, ENTLN. For completeness, the authors should also include GLD360.
- L74: 'excellent' DE. A DE of 38% is far from excellent.
- L166: '...quantities must be considered as global values...': what is meant by 'global' here?
- L190: "...the best data recorded by WWLLN so far was a DE of 31%...". This value is not even in your Table 1.
- L197: '...some of them reporting a DE with errors assumed to be between 80%-90%'. Explain in more detail what you mean.
- L199-201: '...and make very coarse estimations of CC/IC strokes'. I disagree. There has been significant progress over the past decade in detecting IC/CC events using ground-based LF sensors. This is true, for example, for the NLDN and EUCLID networks.
- L248: 'the reference station: the AEMET': AEMET is not an LLS; it is a National Meteorological Service (NMS).
- L256: A small part of Vaisala’s business is devoted to lightning detection. The sentence at L256/257 implies that Vaisala's only/main focus is on lightning detection.
- L279: Meteo-France does not own the sensors in France. Meteorage is the network provider that sells data to Meteo-France.
- L283: Why mention Nunez et al. (2019) when you only use data from 2012? Additionally, this article is only in Spanish and not an international publication.
- L294: '...together with information about the current for the first stroke.' Does this mean only the peak current of the flash is provided and not of all strokes within the flash? This affects Figure 8 and all related discussions about DE variation as a function of peak current.
- Sect. 4.3: The authors now show AEMET observations of three more recent days. Does this mean the authors have contacted AEMET for more recent data, or are these figures simply screenshots from the AEMET website?
- Figure 11: The regions displayed in figures a and b should be exactly the same, which is currently not the case.
- Fig 12: How do you explain the observations in the Northwest and Northeast observed by AEMET that are not present in the WWLLN dataset?
Citation: https://doi.org/10.5194/egusphere-2024-704-RC1 -
AC1: 'Reply on RC1', Enrique Navarro, 17 Jun 2024
reply
Response to comments by reviewer 1.
First of all, the authors would like to thank the reviewer for his valuable comments, which will help us in improving the quality of the manuscript.
Second, despite the comments by other reviewers are still to come, we think that a preliminary response to Reviewer 1 is appropriate due to the open discussion used for the preprint. In this sense, we would like to note that this response must be understood as a preliminary one and changes proposed here may be modified in the final version of the revised manuscript, depending of the rest of reviewers’ suggestions or comments, once they are received.
The reviewer comments are split into general concerns and specific remarks. In the following paragraphs, you can find the response to all the concerns and remarks.
Thank you again for your valuable comments.
RESPONSE TO GENERAL CONCERNS
Concern 1:
- The data spans only a four-month period (January 1 – April 30, 2012) and is outside the typical thunderstorm season, encompassing a total of only 20,651 lightning strokes within the yellow polygon (and drops further to 3389 strokes in the green square, and an even lower amount of detections within the cyan rectangle which is not being mentioned). The rationale provided is that the AEMET dataset for that time was available as open data.
Given that it is now 2024, a more recent dataset would be significantly more valuable. This is not only because it would be more current but also because the AEMET network from 2012 differs from the present-day network. The justification for using this specific dataset, as mentioned in lines 337-339, is insufficient. The authors acknowledge the WWLLN network for providing the lightning location data used in this paper. I don’t see why the authors cannot ask a more recent data set to AEMET. Moreover, the time resolution within this open dataset is only one second, making it unusable for a comparison study.
Response:
We agree with the reviewer that using more recent reference data would be more appropriated, but we have tried to contact AEMET asking for historical data in order to carry the study and we have had no response. It seems that the high investment carried out by national and regional meteorological services may hinder the access by others to their data. Especially difficult is the access to the historical data usually required for scientific purposes. This difficulty may suggest that the existence of other lightning networks, such as WWLLN, with more available historical data, seems then of great interest for the scientific community. It is worth noting that the AEMET and the WWLLN stations have different objectives. The AEMET distribution is intended for a more detailed monitoring of meteorological events happening in Spain, with a dense distribution of stations. As regards the WWLLN, it is a global network intended to provide data to support global studies by using a world-wide low-density distribution of VLF stations. In this sense, the aim of the manuscript is not a complete and deep comparison between AEMET stations and WWLLN stations to determine which of the two networks is the best, but a reasonably good test of the WWLLN performances in order to stablish its capability for detecting lightning activity in the area considered, Spain and surrounding areas, which has not been reported yet in the specialized literature.
Bearing in mind the above-mentioned difficulty in accessing the data from national or regional meteorological services, the relevant question concerning the manuscript under consideration is whether the use of data from 2012 invalidates or reduces the conclusions of the work presented in the manuscript. As mentioned above, there is no study that considers the performances of the WWLLN at western Europe yet, despite this study has been carried out and reported for other areas of the world. Since WWLLN is a global network, we think that the study is pertinent and relevant. The conclusions by comparing data from same time periods, 2012, seems fair and enough to give values of detection efficiency (DE) and location accuracy (LA). The difficulty in measuring lightning activity causes that values already reported in the literature have great error estimations (see Table I in the paper and answer to a comment below in this response). This makes that possible differences that may arise from technical improvements implemented since then in both networks are not so relevant and the conclusions of the presented work, comparing data from WWLLN and AEMET corresponding to 2012, can be considered as still valid in present days and, therefore, we think that the study in this manuscript represents a valid prior calibration stage.
Another important point we would like to note is that in the scientific community, and more specifically in the communities involved in the study of natural phenomena, it is common the study of past phenomena, often occurred several years ago, from a new point of view. In this case, the fact the data may be old might not be as relevant as the data existence and their accessibility. In this sense, the data made public by AEMET are very useful for a study of the present atmospheric phenomena, but we think that WWLLN provides a relatively accessible database of historical past data, which may be useful for scientific studies involving meteorological phenomena during large time spans, as long as these data are duly verified. We think that the work we present contributes to this data calibration study.
In addition to the initial work using 2012 data, a subsequent study of more recent activity has also been addressed and included in the manuscript. The activity considered corresponds to three large storms that generated risks of flooding and hail for the population and mainly putting the Valencian agricultural production at risk: April 2020, August 2021, and August 2022. Once again, we have tried to contact AEMET to ask for recent data but no response from the agency was again the result. The existence of other global networks with more accessible historical data seems again an important benefit. For the study, the screenshots of the available figures in the AEMET website have been used. The results shown in Figures 10 to 12 show a qualitatively good behavior.
As regards the time resolution of 1 s, it is imposed by the available data used in the first study. However, this resolution is similar to that used in some of the previous work presented in Table 1 of the manuscript. For instance, in (Fan et al., 2018) and (Kigotsi et al., 2018), the criterion to set a coincidence is 0.5 s and a distance lower than 50 km, while 0.5 s and 20 km are used in (Abarca et al., 2010). In addition, it is worth noting that the typical duration of a flash is in the order of 0.5 s, while several strokes usually happen during a flash, lasting only about 10-20 ms each stroke. As reported by Jacobson 2006, referenced in the paper, each WWLLN station has a triggering system with and average intertrigger time of 0.2s, which causes that WWLLN detection of a lightning stroke only happens once during a flash and stroke detection becomes effectively a flash detection. Since the typical duration of a flash is around 0.5 s, the resolution of 1 s is enough to compare the lightning activity of AEMET, corresponding to flashes, to the lightning strokes detected by WWLLN, in practice also corresponding to flashes for the reasons stated above.
Concern 2.
-The manuscript is submitted to NHESS, which is dedicated to research on natural hazards and their consequences. Although the current manuscript deals with lightning observations, it is essentially a comparison study. Therefore, the more appropriate journal for such a type of study would be EGU Atmospheric Measurement Techniques, which includes intercomparison research.
Response:
We thank the reviewer for this suggestion. It is certainly right in noting that the paper could have been sent to EGU Atmospheric Measurement Techniques, but we still think that the paper is also appropriated for NHESS. As we mention above, the paper is not exactly about a comparison between AEMET and the WWLLN, since they have different objectives and characteristics, but on studying and presenting the WWLLN performances as a global network. Of course, these performances are assessed by means of a comparison, but the comparison to determine which network is the best is not the final goal of the paper. Bearing this in mind, we think that the content of the paper falls within the second item of the NHESS journal scopes published in the Journal web page (https://www.natural-hazards-and-earth-system-sciences.net/about/aims_and_scope.html):
- “The detection, monitoring, and modelling of natural phenomena, and the integration of measurements and models for the understanding and forecasting of the behaviour and the spatial and temporal evolution of hazardous natural events as well as their consequences”.
The fact that the Editor of NHESS has admitted the manuscript as a preprint seems to support this choice.
RESPONSE TO SPECIFIC REMARKS
Specific remark:
-In the abstract and the text, it should be clarified whether the DE pertains to strokes or flashes.
Response:
Thank you for the remark. The reviewer is right in noting that a certain ambiguity is present in this sense. The WWLLN identifies lightning activity through very low frequency radiation originated by lightning strokes. In this sense, the WWLLN does not detect flashes, but lightning strokes, instead. However, as described in the Jacobson 2006 referenced in the paper, each WWLLN station has a triggering system with and average intertrigger time of 0.2s, which causes that WWLLN detection of a lightning stroke only happens once during a flash and stroke detection becomes effectively a flash detection. This coincidence in the detection of a flash and a particular stroke of that flash may explain that the slight undefinition that the reviewer has found in our paper is also present in other similar papers referenced in the manuscript. In this sense, and only as examples of the different terminology: Jacobson talks about lightning or lightning evens in (Jacobson 2006), Abarca uses the term flashes (Abarca 2010) and Abreu uses the term lightning strokes (Abreu 2010), while Thomson mentions “LIS groups that have a coincident with a WWLLNN or ENTLN stroke” (Thomson 2014).
In any case, the reviewer is right and this aspect will be clarified in the revised manuscript. Specific changes will be addressed and highlighted once we receive the comments and suggestions of the rest of reviewers.
Specific remark:
-In several parts of the text, it is stated that the primary objective of regional and national lightning location networks is to detect CG strokes, with IC events being of lesser importance. I disagree with this. For example, air traffic controllers are very interested in IC activity and usually receive the observations through a LLS network provider.
Response:
The reviewer is right in noting that. Despite the AEMET and other regional and national lightning location networks are devoted to general public and look for CG strokes (these is supported by the references given in the manuscript), but other applications involving IC and also CC events are also relevant, as it is the case of aeronautic applications mentioned by the reviewer. Once the rest of reviewers’ suggestions are received, we will change the sentences in the introductory section to also mention those applications.
Specific remark:
-The introduction mentions another global network, ENTLN. For completeness, the authors should also include GLD360.
Response:
Thank you for the suggestion. The network will be added in the revised version
Specific remark:
- L74: 'excellent' DE. A DE of 38% is far from excellent.
Response:
The reviewer is right. The DE and LA are significantly better than those previously reported in other areas, with DE values below 10%, but “excellent” was inappropriately exaggerated. The sentence in line 74 will be changed in the revised manuscript in the following sense:
“Our work will show that the WWLLN provides values for the detection efficiency (DE) and location accuracy (LA) in the area under study which are higher than those reported up to the moment, with remarkable results for high peak current events.”
Specific remark:
-L166: '...quantities must be considered as global values...': what is meant by 'global' here?
Response:
As explained in the original sentence, the term global refers to the fact that the DE corresponds to the detection of a lightning stroke, regardless of its amplitude. But the detection capability improves with the peak amplitude. A reference to the network performance at different specific amplitudes is also made in the sentence. Maybe using the term global is not a good choice, since the global adjective is also used to define WWLLN as a global network. In this sense, once the rest of comments are received, we will change the term global by total and will slightly change the sentence in the following terms.
“It is worth noting that these two quantities must be considered as total detection values, i.e., they correspond to the detected lightning strokes, independently of their current peak amplitude. More detailed information on the DE values for specific current peak amplitudes can be found in the works referenced in Table 1”
Specific remark:
-L190: "...the best data recorded by WWLLN so far was a DE of 31%...". This value is not even in your Table 1.
Response:
The value corresponds to a reduced area of the larger western hemisphere region considered in the work by (Rudlosky and Shea., 2013), which is included in Table I. The sentence will be changed to clarify this point in the following terms:
“The best data recorded by WWLLN so far had a DE of 31%, obtained in the Pacific Ocean in January 2010, a reduced area of the whole Western Hemisphere region considered in (Rudlosky and Shea., 2013) and shown in Table I.”
Specific remark:
- L197: '...some of them reporting a DE with errors assumed to be between 80%-90%'. Explain in more detail what you mean.
Response:
There is a typing mistake in the original manuscript. It should read “a LA with errors…” instead of “a DE with errors…”. We thank the reviewer for making us note the mistake. Bearing this mistake in mind, the difficulty of measuring lightning activity and knowledge of “ground truth” causes that errors may be comparable to the measured values. As a simple example, let us note that the first reported value of LA in the table by Lay et al in 2004 is of 20.25±13.5 km, i.e., the error is around 67% of the average value. Similarly, the LA value of Abreu et al. 2010 is of 7.24±6.24, which means that the error is around 86% of the value.
Specific remark:
- L199-201: '...and make very coarse estimations of CC/IC strokes'. I disagree. There has been significant progress over the past decade in detecting IC/CC events using ground-based LF sensors. This is true, for example, for the NLDN and EUCLID networks.
Response:
Thanks for your comment. The revised version will eliminate “and make very coarse estimations of CC/IC strokes” from the sentence.
Specific remark:
- L248: 'the reference station: the AEMET': AEMET is not an LLS; it is a National Meteorological Service (NMS).
Response:
Effectively AEMET is not an LLS; it is a National Meteorological Service (NMS), and we will change this in the revised manuscript, it will be changed in the revision to “3. The reference regional lightning detection system of the National Meteorological Service, AEMET”.
Specific remark:
-L256: A small part of Vaisala’s business is devoted to lightning detection. The sentence at L256/257 implies that Vaisala's only/main focus is on lightning detection.
Response:
We will omit that part of the sentence to avoid suggesting that Vaisala’s activity is only devoted to lightning detection. The final sentence will mention the sensors, that they are manufactured by Vaisala and will only add the link to Vaisala website without mentioning any particular activity of Vaisala.
Specific remark:
-L279: Meteo-France does not own the sensors in France. Meteorage is the network provider that sells data to Meteo-France.
Response:
We would like to thank the reviewer for the information. The sentence in line 279 will be changed in the following terms:
“… and from sensors of Meteorage who provides data to the French meteorological service (Météo-France) “
Specific remark:
-L283: Why mention Nunez et al. (2019) when you only use data from 2012? Additionally, this article is only in Spanish and not an international publication.
Response:
We will delete the reference Nunez et al. (2019) in the revised manuscript.
Specific remark:
- L294: '...together with information about the current for the first stroke.' Does this mean only the peak current of the flash is provided and not of all strokes within the flash? This affects Figure 8 and all related discussions about DE variation as a function of peak current.
Response:
As we mention above, a flash event includes several lightning strokes. The first stroke is usually the more energetic one and its peak current provided by AEMET. This peak current from AEMET corresponds to that obtained for a lightning stroke detected by WWLLN and the Figure 8 and subsequent discussion is carried out basing on this comparison.
As we stated in L488-L489, the results of Fig. 8 are very similar to previous works referred to in Table 1 (Abarca et al., 2010; Rodger et al., 2006; Fan et al., 2018), which seems to support the bondage of this procedure and subsequent discussion.
Specific remark:
-Sect. 4.3: The authors now show AEMET observations of three more recent days. Does this mean the authors have contacted AEMET for more recent data, or are these figures simply screenshots from the AEMET website?
Response:
We have tried to contact AEMET for recent data and we have had no response. The Figures from AEMET are screenshots from the AEMET website to make a comparison of recent data. The comparison is then a qualitative one but we think that it is still relevant. We will clarify that the results resented are based on a comparison with screenshots form AEMET.
Specific remark:
-Figure 11: The regions displayed in figures a and b should be exactly the same, which is currently not the case.
Response:
The reviewer is right in noting that. The figures will be corrected.
Specific remark:
-Fig 12: How do you explain the observations in the Northwest and Northeast observed by AEMET that are not present in the WWLLN dataset?
Response:
Thank you for noting that detail which went unnoticed to us. According to the results depicted in Fig. 8, those observations at Northwest and Northeast most likely correspond to low amplitude lightning strokes. We will include a sentence explaining this observation once the rest of reviewers’ comments are received.
Citation: https://doi.org/10.5194/egusphere-2024-704-AC1
-
AC1: 'Reply on RC1', Enrique Navarro, 17 Jun 2024
reply
-
RC2: 'Comment on egusphere-2024-704', Anonymous Referee #2, 19 Jun 2024
reply
see attached pdf; science quality was rated Fair because there may be an error in calculation of DE.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 315 | 49 | 15 | 379 | 12 | 6 |
- HTML: 315
- PDF: 49
- XML: 15
- Total: 379
- BibTeX: 12
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1