Single-site thunderstorm detector using combined electrostatic and radio frequency techniques

Transcription

1 Single-site thunderstorm detector using combined electrostatic and radio frequency techniques A J Bennett 1,2 1 Bristol Industrial and Research Associates Limited (Biral), P O Box 2, Portishead, Bristol, BS20 7JB, United Kingdom 2 Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, United Kingdom alec.bennett@biral.com Abstract Details of a new commercial single-site thunderstorm detector using combined electrostatic and radio frequency techniques are presented, including case studies of its operation. The unit detects and ranges all forms of lightning within approximately 100km using the resultant electrostatic field change, with lightning bearing from the site provided by integrated low frequency radio location capabilities. The use of electrostatics prevents false alarms from radio interference. In addition to lightning location, the instrument detects strong electric field variability and charged precipitation, both of which are used to warn of potential overhead lightning activity. Applications of this new instrument are discussed, including its assistance in the reporting of non-lightning producing Cumulonimbus cloud, which despite being highly significant to aviation is currently not reported automatically, and is challenging to report by human observers during the night. 1. Introduction Lightning is an important observation for nowcasting of severe weather associated with thunderstorms, such as heavy rainfall, strong gusts, hail and the danger from lightning itself. Due to the typical speed of thunderstorm movement it is usually adequate to detect thunderstorms within 60km of a site to provide sufficient warning of their arrival. In most cases, a thunderstorm is already producing lightning by the time it reaches a site, with the distant lightning warning of its presence. However, in a minority of cases the first lightning flash from a developing thunderstorm occurs in the vicinity of the site. In these situations, lightning detection alone offers no warning that the site is about to encounter severe weather. Whilst modern lightning detection technology, such as widespread networks of receivers tuned to the radio pulse from lightning, allow highly accurate lightning locations to be sent to a user in real-time, the problem of nowcasting the first strike remains a challenge. In order to warn of developing thunderstorms in the vicinity of the site, the presence of a Cumulonimbus (Cb) must be detected. This is difficult to achieve automatically or even manually during the night or when the Cb is embedded in low level stratiform cloud. Deep convective cloud such as a Cb can be remotely detected using satellite and radar observations, although in the absence of continuous interpretation by a 1

2 human observer, algorithms for real-time detection of such features are required. Such algorithms are not commonly available to non-specialists. All solutions requiring off-site data collection or interpretation rely on a continuous and reliable data communication link. This may be difficult to achieve for some sites where the reliability of an internet connection is poor, especially during severe weather. A distinguishing feature of deep convective cloud such as a Cumulonimbus is its ability to generate significant electrical charge (Wilson, 1921; Bennett and Harrison, 2007). Whilst the lightning flash is the ultimate observation of this charging ability, charge produced within a developing thunderstorm generates a strong electric field, which can be detected at the ground prior to the onset of lightning activity. A strong atmospheric electric field magnitude has been used to automatically warn of a nearby Cumulonimbus for many decades, usually using an instrument called an electric field mill (e.g. Chubb, 1990). Whilst well suited for warning of strong electric fields, such a device usually needs to be combined with an alternative method of lightning detection so that the system can reliably detect distant lightning, or provide a direction to the thunderstorm. A thunderstorm detector has recently been developed by Biral, a UK-based meteorological sensor manufacturer, with its commercial launch in The objective of this new design was to address the challenges traditionally encountered with single site lightning detectors, such as comparatively high false alarms and range uncertainty. Furthermore, the ability to combine lightning location and overhead Cumulonimbus detection in a single instrument was considered a key benefit not adequately addressed by current meteorological technology. A picture of the Biral thunderstorm detector, named the BTD-300, is shown as Figure 1. The reason for its unusually shaped antennas is described in the following section

3 Figure 1: The Biral thunderstorm detector, BTD Lightning detection and ranging The electromagnetic emission of lightning is strongest in the radio spectrum. Lightning is conventionally located from a single site by interpretation of this radio signal. Whilst sensitive, discrimination between lightning and radio signals resulting from human activity present a significant challenge. Determinationn of range is also difficult to achieve with reasonable uncertainty due to the inherent variability in source strength and waveform modificatio on as it propagates through the atmosphere. An alternative e to electromagnetic (radio) detection is monitoring of the slow variation (< 50Hz) in electrostatic field of the atmosphere, resulting from charge neutralisation inside the cloud by the lightning flash. This step change in the electrostatic field of the thunderstorm has the advantage that its magnitude is much more closely related to range than for radio signals, even after variability in lightning strength is considered. There are also considerably less sources of interference for electrostatic field measurements, since the charge required to generate an appreciable electrostatic field in the atmosphere more than a few tens of metres away are substantial and not usually producedd by non-thunderstorm processes. To allow the antenna to operate under the strong electric field below thunderstorms, the antenna must not initiate corona discharge. Corona discharge is the electrical breakdown of air adjacent to a conductor, which will cause currents to flow through the antenna. To avoid this, the antennas on the BTD-3000 (Figure 1) are smooth and rounded, since the electric field is enhanced near to sharp conductors. 3

4 The strength of the electric field change produced by lightning is proportional to the inverse of the distance cubed, providing the vertical extent of the lightning channel is small compared to the distance to the receiver, so a small dipole can be assumed. The change in electric field ( E S ) produced by a vertical lightning channel of height H over a flat conducting surface and charge neutralisation Q at a distance D from the receiver can be calculated as follows: E S QH D H 2 (1), where ε 0 is the permittivity of free space. Use of equation (1) and typical values of the magnitude and height of charge neutralisation by lightning enable the range to a lightning flash to be determined from the electric field change measured by the BTD Despite the requirement for charge magnitude and height to be estimated, it can be seen by inspection of equation (1) that the distance term (D) dominates the sensitivity due to the inverse-cubed relationship. The change of electric field with time induces a current on an exposed conductor according to equation (2): I ES A (2), t 0 where I is the induced (displacement) current on the conductor, A is the surface area of the conductor exposed to electric field E S change in time t. An additional consideration would be the alteration of the electric field in the immediate vicinity of the conductor, as a result of the shape of the conductor and nearby grounded objects and their elevation above the surface. It is this induced current which is the fundamental measured parameter of the BTD-300. The duration of a lightning flash is typically a few tenths of a second, with a median value of around 200ms. In order to capture the signal shape, including peak amplitude, observations of the electrostatic field must be made at least once every 10ms, i.e. 100Hz sampling rate. An example of the atmospheric electric field change associated with lightning measured by a BTD-300 at 100Hz is shown in Figure 2. The three lines on the graph represent the current induced on the three antennas of the BTD-300, as shown in Figure 1. It can be seen that the top antenna (black line) receives a stronger signal than the lower antennas. The relative difference in signal amplitude between the three antennas is used by the BTD-300 to aid discrimination between lightning and non-lightning signals, thereby reducing false alarms. Further information on this technique is reported by Bennett (2013). 4

5 Figure 2: Example of an atmospheric electric field change associated with distant lightning, measured by a BTD-300. The three lines on the graph correspond to current inducedd on the upper (black), middle (orange) and lower (blue) antennas. Once lightning is detected by the electrostatic field change, the range is calculated based on the magnitude of the change associated with the complete flash duration. The BTD-300 uses electrostatic antennas of approximately 0.1m 2 each, enabling a detectable current to be induced by lightning up to 100km away. Whilst there are very few anthropogenic sources of strong electric field change to be confused with lightning signals, charged precipitation represents a natural signal with similar characteristics to distant lightning. Unlike lightning however, the source of the current on the antennas is from direct contact or very nearby (few centimetres) movement of a charged precipitation particle (hydrometeor). Since the three antennas of the BTD-300 are separated by distances of more than a few centimetres, a precipitation particle will only affect one antenna at a time. It is this difference that is used by the BTD-30signals, substantially reducing the false alarm rate. The to differentiate between strongly charged precipitation and lightning similarity in shape and amplitude between lightning and charged hailstones falling on the BTD-300 is shown in Figure 3. In this graph the negative pulses are from individual hailstones passing close to one of the BTD-300 antennas. The only pulse which is consistent in shape and polarity between all three antennas is at 11:26:53.9 UTC. This cannot be produced by a charge at close proximity such as a hailstone, but is instead generated by a distant lightning flash. 5

6 Figure 3: Induced current on the threee BTD-300 antennas as hailstoness fall nearby. The three lines on the graph correspond to current induced on the upper (black), middle (orange) and lower (blue) antennas. A single distant lightning flash is also evident at 11:26:53.9 UTC. Measurement of electrostatic field changes using three co-located antennas offers a reliablee method of lightningg detection and ranging. Due to the necessity for electric field lines to be perpendicular to the surface of conductors such as the ground, it is not possible to derive an unambiguous estimate of the direction of lightning from a single measurement of electrostatic field change. The method used by the BTD-300 for lightning direction finding is described in the following section. 3. Lightning direction finding The radio emission from lightning can be used to find its direction by a technique called magnetic direction finding (MDF). This techniquee requires three co-located horizontal magnetic field component of the radio signal respectively, and a third one which antennas, two of which are sensitive to the north-south and east-west measures the vertical electric field component. The relative magnitude and polarities of these radio wave components determines the direction of the source (Rafalsky et al. 1995). Compared to the electrostatic antennas, the size of the MDF antennas are small, with the electric field antennaa a thin electrode of less than 10 cm long and can be placed inside a small non-conducting box. An image of the MDF module on the BTD-300 is shown as Figure 4. 6

7 Figure 4: The lightning magnetic direction finding module of the BTD-300 thunderstorm detector. The MDF module operates in the Low Frequency (LF) portion of the radio spectrum, with a peak sensitivity at ~80kHz. The selection of this frequency offers a compromise between the ~10kHz peak amplitude from cloud-to-ground of intra-cloud lightning. Despitee the physically small antenna sizes, the MDF module can detect lightning at return strokess and the higher frequencies more characteristic more than 300km range, so the limiting factor for the reporting range of the BTD- stroke 300 is the ~100km range capability of the electrostatic antennas. An example of the output from the MDF antennas associated with a lightningg 60km to the NE is shown in Figure 5. Unlike the electrostatic signals for a complete lightning flash, the radio pulses associated with individual lightning strokes are short, with a duration of ~10 microseconds. Therefore, considerably higher sampling rates are required, with sampling by the MDF module being 800kHz, compared to only 100Hz for the electrostatic antennas. Since a single lightning flash may comprise of several individual strokes, the radio signal from the stroke with the highest amplitude is used to determine its direction. Due to the prevalence of artificial sources of radio, a lightning flash is only reported by the BTD-300 if detectedd by the electrostati ic antennas. Once a flash is detected electrostatically, the sensor polls its MDF module to determine the direction of the lightning signal. In this way, the benefits of low false alarms and more precise rangingg from electrostatic detection are preserved, whilst also permitting a lightning direction to be found using conventional radio techniques for single receivers. 7

8 Figure 5: Raw antenna output from the magnetic direction finding module of the BTD-300 sampled at approximately 800kHz, showing current pulses from a lightning stroke on N-S and E-W magnetic field components and the omnidirectional electric field component. The combination of electrostatic detection and ranging with radio direction finding allows the location of all lightning flashes, irrespective of type (both cloud-to-ground and intra-cloud) to be determined by a single-site detector such as the BTD-300, during local to a range of approximately 100km. An examplee of the BTD-300 output thunderstorm activity near Bristol, UK, on 17 June 2016 is shown in Figure 6. In this example, a thunderstorm is shown slowly tracking from 12km NW to 15km SW of the BTD-300, with other less active thunderstorm cells apparent further to the north and west over Wales. This slow moving thunderstorm produced intense rainfall in the Bristol Channel. A stationary line of thunderstorm activity was also detected 20-60km to the SE of the sensor over east Somerset, with the resultant high rainfall accumulation causing localised flooding. 8

9 Figure 6: Lightning flash locations reported by a BTD-300 (located at the central red cross on the map) near Bristol, in south-west England. Flash locations are filled circles, with their colour indicative of the time of occurrence. The grey concentric circles represent range from the detector, at 10 km intervals. 4. Detection of an overhead Cumulonimbus Not all Cumulonimbus clouds produce lightning, although their presence at a site indicates an increased risk of severee weather, such as flash flooding, blizzards or hail. Identification of this cloud type is particularly important for aviation, since it can produce severe turbulence, airframe icing and microbursts. Aerodromes also need to report when a Cb is overhead as significant weather, due to these hazards. As mentioned in the introduction, a distinct feature of these clouds is their strong charge, allowing their overhead presence to be detected at the ground by an increased atmospheric electric field magnitude. Except during lightning events, the electrostatic field associated with a Cb varies over a timescale of minutes. The BTD- filter. 300 cannot directly measure this steady-statee field due to a 1Hz high pass Instead, a strong electric field is detected by measuring the charge on individual hydrometeors as they fall on the antennas, which is proportional to the atmospheric electricc field. When no precipitation is falling at the site, a strong field is detected through the significant increase in ion production near the surface from corona discharge initiated by the strong field magnitude. The ion clouds are transported past the 9

10 antennas in turbulent eddies, wheree they induce currents. The characteristics of these signals (such as good correlation between all the electrostatic antennas) are used to infer the existence of corona ion clouds and therefore a strong atmospheric electricc field. When an overhead Cb is strongly charged, it is common for signals from both charged precipitation and corona ions to be detected, such as shown in Figure 7. In this example, a strongly varying current is induced on the top antenna of a BTD-300 approximately two minutes before an overhead ( <9 km) lightning flash. Although short, this 2 minute warning was especially useful since it was the first lightning flash of the storm, providing an alert thatt a Cb was overhead, with associated increase of a nearby lightning risk. Figure 7: Strong variability of current induced on the top antenna of a BTD-300 before and after a nearby lightning flash (the large bipolar spike at 11:19:10 UTC). Warning of potential nearby lightningg activity is of particular use if the lead time is sufficient to allow mitigation action such as moving people to a substantial building or stopping aircraft refuelling. The warning lead times to nearby lightning flashes provided by charged rain or corona ion detection from a BTD-300 is shown in Figure 8. The BTD-300 was located in Bristol, UK, and the time period for this assessment was between 1 January and 31 July From this figure is can be seen that the combination of charged rain and corona ion detection provided between 1-30 minutes warning of nearby (<20 km) lightning flashes. Although short compared to that usually provided by distant lightning, the few minutes would be sufficient to take immediate action to safeguard people, such as golfers or airport ground staff to move into a nearby shelter. Moreover, 2 of the 14 thunderstormss which produced nearby lightning during this period were only warned about in advance by charged rain or strong electric field variability, with no prior lightning detected. Similar results were also reported by Bennett (2016) analysing BTD-300 output over a 17 month 10

11 period. Whilst these parameters are effective at detecting an overhead Cb, with a considerably low false alarm rate, it must be emphasised that not all will go on to produce lightning, remaining instead as shower clouds. As a consequence, the false alarm rate from using strong atmospheric electric field magnitudes as an indicator of potential imminent lightning activity is relatively high, at 30-80%, (Bennett, 2016). The high false alarm rate should be considered in relation to the benefit that an early warning of increased lightning risk has for the site, as well as the non-lightning hazards presented by a Cb. For comparison, during the false alarm rate for tornado warnings in the USA issued by the National Weather Service was 76% (NWS, 2011). Figure 8: Warning lead times from a BTD-300 for nearby lightning flashes, provided by charged rain (yellow) and strong electric field variability due to corona ions (orange). The assessment was made at Bristol, UK, between 1 January and 31 July Conclusions The Biral BTD-300 is a single-site thunderstorm detector combining electrostatic and radio detection techniques, capable of locating lightning up to approximately 100km and warning of on overhead Cumulonimbus several minutes beforee the first nearby lightning flash. Distant lightning is usually an effective provider of early warning of an approaching thunderstorm. However, warnings derived from in situ measurements of atmospheric electrical parameters such as charged precipitation or strong electric field variability are particularly important in the minority case when thunderstorms develop over a site, since they can provide early warning before the first lightning flash is generated. Even if the overhead Cumulonimbus does not produce lightning, its automatic detection aids the reporting of this significant and potentially hazardous weather phenomenon by sites such as airfields and heliports. 11

12 References Bennett, A. J Warning of imminent lightning using single-site meteorological observations. Weather, In Press. Bennett, A.J Identification and ranging of lightning flashes using co-located antennas of different geometry. Measurement Science and Technology, 24, Bennett, A. J. and Harrison, R. G Atmospheric electricity in different weather conditions. Weather, 62(10), Chubb, J.N., Two new designs of field mill type fieldmeters not requiring earthing of rotating chopper. IEEE Transactions on Industry Applications, 26(6), NWS, NWS Central Region Service Assessment Joplin, Missouri, Tornado May 22, U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, Central Region Headquarters, Kansas City, MO, July Rafalsky, V.A., Nickolaenko, A.P., Shvets, A.V. and Hayakawa, M., Location of lightning discharges from a single station. Journal of Geophysical Research, 100(D10), Wilson, C.T.R., Investigations on lightning discharges and on the electric field of thunderstorms. Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 221,