JP2009085965A - Lightning detection system and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect information pieces about both lightning discharges in a CG and an IC, using a plurality of remote-programmable sensors (RPS) arranged in different geographical positions, to classify them, to package them, to transmit them by compressed forms, to collect them in a central analyzer position, to expand the information pieces to determine a position of the lightning discharges, a magnitude thereof and a generation time thereof, in a central analyzer, and to correlate them each other. <P>SOLUTION: The present invention discloses a lightning detecting and data acquiring system. The plurality of remote programmable sensors is used to detect cloud-to-ground and IC lightning strokes. Each analog expression of the lightning stroke is converted into a digital signal. The digital signals are classified according to user changeable criteria. The classified digital signals are compressed and optionally decimalized. The compressed information is transmitted to a central position, and the compressed information is expanded in the central position, to be used for correlating the position, the magnitude and a moving route of the detected lightning stroke. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a lightning detection system and method, and more particularly to a programmable system that provides continuous lightning detection and enables user-selectable evaluation criteria.

  The lightning detection system and data acquisition system are used to detect the occurrence of lightning discharge, determine the location of the lightning discharge, and collect other data about the discharge. In the conventional lightning detection system, a plurality of sensors are arranged several tens to several hundreds kilometers apart to remotely detect the electromagnetic field of lightning discharge. Such discharges can occur between the cloud and the ground (CG) or in the cloud (IC). Information from the sensor is transmitted to the central station, where the sensor data is analyzed. Typically, the discharge occurrence time and position are determined from data provided by multiple sensors.

  The lightning detection and data acquisition system remote sensors typically detect both electromagnetic fields in CG and IC lightning flashes consisting of many discharges. It is often important to be able to distinguish between two types of flash. For this purpose, remote sensors often face low frequency (LF) and very low frequency (VLF) emissions from lightning discharges. Since the antenna is inherently responsive to the time derivative of the electrical or magnetic discharge field, the electrical signal generated by the LF and VLF (LF / VLF) detectors typically produces a waveform that represents the electrical or magnetic discharge field. To be integrated before analysis. Analyzing a signal representing either an electric field or a magnetic field that identifies the discharge of CG and IC will be referred to as performing waveform analysis. A well-known method of identifying lightning signals in both the LF and VLF ranges is to examine the time elapsed from the peak in the representative signal to the instant of crossing the zero amplitude reference point. This is called a peak-to-zero (PTZ) analysis method. A relatively short PTZ time is a good indicator that an IC discharge has occurred. Another well-known identification method is called bipolar test, in which the representative signal is the first peak and the next peak of opposite polarity that is greater than a predetermined percentage value of the first peak. Find out about. Such a method is another good indicator of IC discharge. Yet another test for IC discharge is the presence or absence of a next peak that is greater than the initial peak of the same polarity in this representative signal. This is to be expected based on the fact that some IC discharges have a number of small and fast preceding electromagnetic pulses before a subsequent larger and slower pulse. If there is no such criterion indicating that the discharge is an IC discharge, it is usually assumed to be a CG discharge. Even applying all established criteria to distinguish between CG and IC events, some events are still misclassified.

  Another method of lightning detection is to monitor very high frequency (VHF) radiation from lightning discharges. However, the VHF detection system must be able to process information at very high data rates because VHF pulse radiation in IC lightning occurs about a tenth of a millisecond apart. In addition, the VHF system can only detect lightning events that have a direct line of sight to the sensor. Such a system is currently used by NASA at the Kennedy Space Center in Florida. However, this system is further limited to a direct line of sight between the sensor and the central analyzer because it uses a real-time microwave communication system. In addition, the VHF system used by NASA has proven to be expensive to install and maintain.

  Previous lightning detection and data acquisition systems that detect low-frequency electromagnetic field signals were designed around a combination of time-of-arrival (TOA) and magnetic direction finding and time-domain field waveform analysis. In most of these systems, the sensors are primarily analog devices. Using an analog device for the lightning detection sensor requires the use of a “track and hold” circuit to detect moderate events, capture this representative signal, and perform waveform analysis there. Due to the accumulation of delay periods in these “track and hold” circuits, these sensors have a large “rearm” time or “dead time” time during which they cannot record subsequent lightning events. Even newer lightning detection and data acquisition systems that are substantially digital have some dead time. For example, sensors in some such systems have a “dead time” of 5-10 milliseconds, and most current digital sensors have a “dead time” of up to 1 millisecond. The latter can only detect a limited part of the IC lightning discharge. This is due in part to the fact that some IC lightning discharges occur in 1 millisecond. However, CG lightning flashes tend to have less discharge with a relatively long period between individual discharges. If previous generation sensors were designed to monitor CG and IC electromagnetic field signals, a significant portion of the time was taken up in handling IC discharge events, at the expense of event recording. Other aspects regarding sensor dead time and TOA location methods become uncertain when multiple remote sensors assume that they will respond to the same IC lightning event. Remote small-amplitude events are difficult to detect when the electromagnetic waves travel long distances on the earth due to the attenuation they suffer. If different sensors generate arrival time information from different events, this calculated discharge location will have significant errors.

  Analog sensors operating at LF / VLF frequencies are difficult to adjust for both CG and IC lightning discharges. The median amplitude of the CG field signal relates to an order of magnitude greater than the median field signal amplitude of the IC. By optimizing the gain of one of these sensors for the detection of IC events, the sensor is often saturated with the much greater energy of a nearby CG lightning discharge. Therefore, adjusting the gain to accommodate both types of electromagnetic field signals typically reduces the ability of the sensor to detect IC events. Since distant IC lightning discharges are attenuated by propagating through the ground, they are difficult to distinguish from background environmental noise.

  In order for the lightning detection system to provide useful information in a timely manner, there must be a way to transmit sensor information to a central location. This central location must collect information from a number of remote sensors, and this information is then correlated to establish the location, magnitude, and time of occurrence of the lightning discharge. Existing detection systems generally have a narrow bandwidth communication system that limits the amount of information that a sensor can send to a central analyzer. In many existing lightning detection networks, the sensor is connected to the central office by a low speed telephone modem, typically 2400-9600 bit per second. In the past, this communication limitation was not overly deterministic because the large dead time of previous generation analog sensors limited the amount of information that could be collected and sent to the central analyzer.

  When sensor information arrives at the central office, it must be analyzed. Information from each sensor is compared with incoming information from other sensors. This comparison process attempts to find the corresponding data to determine the location, magnitude and time of occurrence of the lightning discharge. However, current correlation techniques are insufficient to process large amounts of information if the time between discharges is two orders of magnitude shorter than the travel time between sensors. In practice, if the lightning detection system uses advanced techniques to send and receive increased amounts of information, current central analyzers cannot efficiently process the information with current correlation techniques. Will.

  The state of the art of lightning detection systems and data acquisition systems is generally manifested in part by several patents. First, U.S. Pat. Nos. 4,1985,599 and 4,245,190 to Krider et al. Describe a network of gated broadband magnetic direction finding sensors. These sensors are sensitive to return strokes in CG lightning flashes. In US Pat. No. 4,1985,599, identification and classification is accomplished by examining the shape of the time domain field waveform. A short rise time (threshold to peak time) occurs as a result of the representative signal being placed in the analog track and hold circuit while further analysis is being performed. These sensors are primarily designated by the CG discharge of interest. Any IC lightning discharge detected is abandoned. However, both CG and IC events that meet a short rise time criterion and a simple test of event duration result in a significant amount of sensor dead time.

  Second, Bent et al. U.S. Pat. Nos. 4,543,580 and 4,792,806 disclose a network of sensors that measure the TOA of electromagnetic field signals and use this information to locate lightning. These sensors do not distinguish between IC discharges and CG discharges. However, these sensors suffer from dead time problems similar to the magnetic direction sensor of Krider's patent. If multiple IC discharge pulses occur in a short time, there is no guarantee that multiple sensors will respond to the same IC discharge event.

  Another patent of interest is Markson et al., US Pat. No. 6,246,367, in which the lightning detection system uses an analog-to-digital converter (ADC) and is representative of Provides continuous processing of electromagnetic field signals. This eliminates the dead time problem inherent in previous generation sensors. Markson describes using a bipolar comparator to distinguish between positive and negative polarity versions of a particular pulse that is inferred to be the first broadband radiation pulse in a CG or IC flash. Markson also uses a data correlation process and an arrival time difference location method. Markson's approach obviously uses a high-pass filter to block most of the low frequency components of a typical electromagnetic field signal, so this is not always useful for detecting the initial pulse in the flash. The limitations of the Markson patent are the specific use of the HF frequency range and the detection and processing of only the first pulse in each flash.

  Accordingly, there has been a need for improvements in lightning detection systems and data acquisition systems in several respects. First, an improved signal conditioning method is needed. CG events are typically an order of magnitude larger than IC events at LF due to the channel length and amount of current that flows during a CG return lightning strike. As mentioned above, increasing the gain or equally reducing the event threshold results in a CG event saturating the analog detection and evaluation system, or a significant amount of noise pickup. Increasing gain or equally reducing the event threshold results in poor detection masking IC events. While removing unwanted noise components, the effect of this magnitude difference between CG and IC signals needs to be reduced. An interesting aspect of both electric and magnetic field antennas is that they generate signals that are proportional to the time derivative of the electromagnetic field they are detecting. These differential antennas actually reduce the magnitude imbalance between the differentiated representative signals of IC and CG. However, current generation sensors always impose an integration method that converts the differentiated electromagnetic field signal into an integrated signal representing the electromagnetic field without using the fact that the antenna itself reduces the dynamic range requirements. In addition, there is a need for improved classification methods for identifying lightning types.

  Another desire in the industry is to allow remote sensors to be programmed with new or different waveform analysis techniques. There is also a need for improved data comparison and data decimation techniques that accommodate more IC and CG information. Thus, new data correlation techniques are needed to handle increased information processing rates. These correlation techniques need to handle both arrival time and direction information.

  Therefore, a new method of signal conditioning, a user-changeable system for event classification, a new method of data comparison, and CG and IC events are efficiently detected and their location, size and time of occurrence are determined. There is a need for a complete lightning detection and information acquisition system combined with new data correlation techniques.

SUMMARY OF THE INVENTION The present invention uses a plurality of remote programmable sensors (RPS) located at different geographical locations to detect and classify information regarding both CG and IC lightning discharges. Fill by packaging and transmitting in compressed form. Information is collected at a central analyzer location where the information is stretched and correlated to determine the location, magnitude and time of occurrence of lightning discharge. An antenna designed to detect an electromagnetic field signal from a lightning discharge and generate a representative electromagnetic field signal of a derivative is used. The derivative signal has the advantage of reducing the amplitude imbalance between the CG and IC electromagnetic field signals. The filter is used to increase the signal-to-noise ratio by passing through the low frequency part of the differential signal and discarding the frequency noise without integrating the fundamental component of the signal. Non-linear amplification further reduces the amplitude imbalance between CG and IC signals by providing greater amplification for smaller amplitude signals. The amplified signal is then processed by the ADC to convert the amplified differential signal into a digital representation signal. This conversion allows the signal to be processed and stored digitally. The digital representation signal is then integrated by a digital processor to provide a signal representative of the electric or magnetic field. The digitally differentiated electromagnetic field signal and the digital representation signal of the electromagnetic field itself are used by the digital processor to classify lightning events as either CG or IC events. Analog-to-digital conversion combined with digital storage allows continuous detection and evaluation of lightning discharge, which eliminates the “dead time” inherent in previous generation lightning detection systems.

  The present invention uses a new data compression process to transmit data over a narrowband communication channel. Multiple digital signal pulse representations of lightning discharges are grouped together in a pulse train. The largest pulse is designated as the reference pulse and its magnitude, time and direction (if available) are included in the data record. Other pulses of the pulse train are represented by an amplitude that is a proportion of the reference pulse and a time stamp relative to the time of the previous or subsequent pulse. This greatly reduces the information that must be transmitted in order to accurately define every pulse in the pulse train. If the amount of information transmitted still exceeds the bandwidth of the associated communication channel, RPS sensors in the lightning detection system can be programmed to transmit a synchronized portion of the information, and all sensors Information about the same lightning event can be reported.

  The information is unpacked when received by the central analyzer and the original pulse amplitude, time and direction (if available) information is recovered. Unpacked pulse information is used to correlate lightning strike information from multiple sensor locations. This information is used to determine the magnitude, location and time of occurrence of the lightning discharge.

  Accordingly, it is a primary object of the present invention to provide a novel and improved lightning detection system and data acquisition system and method.

  Another object of the present invention is to provide a lightning detection system and data acquisition system and method with improved ability to identify CG and IC lightning events.

  It is yet another object of the present invention to provide a lightning detection system and data acquisition system and method that reduces amplitude imbalance between typical electromagnetic field signals of CG and IC lightning.

  It is an additional object of the present invention to provide a lightning detection system and data acquisition system and method for continuously detecting and processing electromagnetic field signals generated by lightning discharge.

Still another object of the present invention is to determine the location, size and time of occurrence of a lightning strike.
A lightning detection system and data acquisition system and method for improved correlation of information from a plurality of remote programmable sensors.

  Still another object of the present invention is to provide a lightning detection system and a data acquisition system and method in which the arrangement of sensors can be set or changed by remote access.

  The above and other objects, features and advantages of the present invention will be more readily understood upon consideration of the following detailed description of the present invention taken in conjunction with the accompanying drawings.

  Referring to FIG. 1, a preferred embodiment of a lightning detection system and data acquisition system 10 is illustrated. Remote programmable sensors (RPS) 12 are distributed tens to hundreds of kilometers apart. These remote sensors are used to detect the electromagnetic field generated by the lightning discharge from the cloud 22 as either CG lightning 14 or IC lightning 16. The communication link 18 enables the remote sensor 12 to send information to the central analyzer 22 where the position, magnitude and time of occurrence of lightning discharge is determined.

  The CG lightning discharge is typically 10 times greater in amplitude than the IC event in the VLF / LF frequency range. In order to prevent saturation of the analog components of the remote sensor 12, means are provided to reduce the amplitude imbalance between the CG representative signal and the IC representative signal. The sensor 12 also has means for increasing the signal to noise ratio of these representative signals. In addition, the sensor 12 converts the analog representative signal into a digital signal, which is then classified as either CG or IC lightning. Once the signals are classified, the sensor determines whether the group of signals is close together in time enough to be considered a train of pulses. For such a pulse train, the sensor packages the signal information into compressed data and transmits these data words to the central analyzer 20. The separated pulses are transmitted alone with a richer set of characterizing features. If the communication link is insufficient to process all the information, the remote sensor 12 decimates the information and sends only the synchronized portion of the information.

  The central analyzer 20 is used to receive the data word sent from the remote sensor 12 and expand the word to obtain lightning discharge information. Using correlation techniques, information from the plurality of sensors 12 is used to determine the size, position and time of occurrence of the lightning discharges 14,16.

  A lightning flash occurs between charge accumulations of opposite polarity. A lightning flash begins with a small breakdown because the air during charge accumulation is ionized and forms a conductive channel. In a CG flash, when a channel is formed from the cloud to the ground, a large amount of electricity flows between the cloud and the ground. The discharge that generates these large currents is called a return stroke. A typical CG lightning flash has four return strokes. These lightning strikes are typically tens of milliseconds apart. A waveform 30 of the electromagnetic field generated by the CG lightning strike is shown in FIG. This figure shows a first negative peak 32, a first negative trough 34, a second negative peak 36, a zero crossing point 38, a first positive peak 40, a first positive trough 42, and a second positive peak 44. Is illustrated. The second signal 46 describes an electrical signal representing the wave of the electromagnetic field after detection by the antenna 12 according to the time derivative of the electromagnetic field.

  The number of discharges in the IC flash is about 10 times greater than the number of lightning strikes in the CG flash. On the other hand, the median amplitude of the electromagnetic field produced by CG lightning strike is about 10 times greater than that produced by IC discharge. The period between pulses in the IC discharge is much smaller than for CG lightning. The largest amplitude electromagnetic field pulses generally occur in the middle of these pulse trains. The waveform 50 of the electromagnetic field generated by the IC discharge is shown in FIG. This figure illustrates an electromagnetic field pulse train consisting of a number of prominent pulses 52 and several small to medium pulses 54. The second signal waveform 56 describes the electrical representation signal of the time derivative of the electromagnetic field after detection by the (differential) antenna 12.

  FIG. 4 shows an empirically determined cumulative distribution of signal amplitudes normalized for IC and CG discharge ranges. Range normalization results in signal amplitude values that are independent of the actual distance between the sensor and the discharge. A first curve 70 in FIG. 4 is an expression obtained by normalizing the range of the CG event. The region is the signal intensity in LLP units, the range of which is normalized to 100 km from the sensor position. Referring to the first curve 70, 50 percent of CG lightning events have an amplitude greater than 120 LLP units, and 50 percent are less than 120 LLP units. About 80 percent of CG events have an amplitude of less than 180 LLP units and 20 percent are greater than 180 LLP units. The second curve 72 represents a distribution in which a large and prominent discharge range in the IC flash is normalized, and the third curve 74 illustrates a distribution in which all IC discharge ranges are normalized. About 70 percent of all IC events are less than 1 LLP. This amplitude imbalance requires a novel approach to lightning detection and processing if both CG and IC events are to be detected by the same sensor in the LF / VLF frequency range.

  Turning to FIG. 5, a block diagram illustrates functional aspects of a lightning detection system and data acquisition system according to the present invention. The differential antenna 92 is used to detect either the electric or magnetic field generated by the lightning discharge. The antenna 92 outputs an analog representation signal (electrical detection signal) of the detected electromagnetic field wave and sends it to the remote programmable sensor 94. The first stage of the remote sensor 94 is used to reduce amplitude imbalance between the CG representative signal and the IC representative signal, reduce noise, and convert the conditioned electrical detection signal to a digital representation. A signal adjustment circuit 96 is provided. Once the representative signal is adjusted, it is passed to the event classification stage 98, where the digital representative signal is evaluated and any type of event (either CG or IC discharge) produces an electromagnetic field. Decide. The remote sensor can be programmed based on user-selectable criteria that result in only the signal of interest being accumulated. The accumulated signal of interest is processed by the data compression software 100. If necessary, these are also processed by the data decimation stage 102 and transmitted to the central analyzer 104 using any digital data transmission means. In the central analyzer, the position determination unit 108 uses the data decompression unit 105 and the data correlation unit 106 in the subsequent stage to determine the size, position, and occurrence time of the lightning event.

  Referring to FIG. 6, a structural block diagram of a preferred embodiment of the remote programmable sensor 110 is illustrated. The analog front end 112 receives an analog representative signal from the electromagnetic antenna, filters the representative signal using a filter and amplification component 124, and uses the crosspoint switch 126 to filter the representative signal. The data is passed to the ADC 114 and the amplification comparator 128. A comparator 128 is used to determine whether the amplitude of the analog representative signal exceeds a previously determined value and to indicate the start of a “pulse”. In response to the determination that the pulse has begun, the global positioning system (GPS) device 130 provides a time stamp that is stored in a time tag first in first out (FIFO) 132 in the field programmable gate array (FPGA) 116. used. The ADC 114 performs continuous digitization of the signal provided by the crosspoint switch 126. The ADC generates a digital signal with a 12-bit resolution sampled at 20 MHz. The digitized signal is sent to the digital adder 134 in the FPGA 116 described above. Adder 134 is used to add a group of four digital samples and generate a 5 MHz sample with 14-bit resolution. These 14-bit digital samples are placed in a signal FIFO 136 in the FPGA 116. The clock signal provided by the FPGA is used to control the flow of digital samples from the signal FIFO 136 to the ring buffer 138 present in the synchronous dynamic random access memory device (SD-RAM) 120. Access to the ring buffer 138 occurs by a dynamic memory access controller (DMA) 140 that is part of a digital signal processor (DSP) 118. The DMA controller 140 also transfers the event time stamp from the time tag FIFO 132 in the FPGA 116 to the time tag buffer 142 in the SD-RAM 120. A central processor unit (CPU) 144 provided in the DSP 118 is used to evaluate the data stored in the ring buffer 138 and the time stamp information stored in the time tag buffer 142. Data representing the signal of interest is placed in a result buffer 146 in the SD-RAM 120. The information then proceeds through the DSP-PC interface 148 to the host personal computer 122 where it is packaged for transmission.

  Signal conditioning and classification aspects of the remote programmable sensor 110 are addressed thereafter and are illustrated by the flowcharts in FIGS. 7, 8, and 9 and FIGS. 10A-10H.

Analog Processing The basic components of the antenna filter network 300 are shown in FIG. The antenna 301 provides the differential signal to a low-pass filter 302 (four poles or more) and a high-pass filter 304. The filter is followed by an optical nonlinear amplification stage 306. Both electric and magnetic field antennas generate signals that depend on the time derivative of the electromagnetic field signal they monitor. Therefore, this antenna 301 is called a differential antenna. This differential characteristic of antenna 301 is important because it has been discovered that differential electromagnetic signals reduce the amplitude imbalance between CG and IC representative signals by a factor of two to four. Previous generation lightning detection systems did not use this antenna characteristic. In practice, most lightning detection systems integrate the antenna output to get a true representation of the electromagnetic field.

  A 4-pole low pass filter 302 is used to provide a good transient response to the differential signal. The output of the low pass filter 302 is sent to a high pass filter 304 having a cutoff frequency of about 300 hertz. The purpose of this filter is to remove any possibly artificial signals such as 50/60 Hz power line noise. No filtering is used in the frequency range between these two filters. This preserves the differential characteristics of the signal throughout this band where the signal of interest may exist. This adjustment of the filter effectively forms an analog leak integrator that can be adjusted to pass frequencies below 0.5-1 MHz without being integrated. An optional component of the present invention is the use of a non-linear amplifier 306 that further reduces the amplitude imbalance between the CG and IC representative signals. A first curve 310 in FIG. 8 shows the frequency response of a leaky filter network including a differential antenna in the preferred embodiment. The second curve 312 illustrates the frequency response of a leaky filter network tuned to remove artificial radio frequency noise whose time constant is excessive.

  The non-linear amplifier 306 of the preferred embodiment amplifies a small amplitude signal more disproportionately than a large amplitude signal. Two types of nonlinear amplifiers suitable for this application are logarithmic amplifiers and piecewise linear amplifiers. Both of these types of amplifiers can reduce the amplitude imbalance between CG and IC signals by 12-24 dB. FIG. 9 illustrates the input-output characteristics of these nonlinear amplifiers. The first curve 330 shows the logarithmic amplifier response, and the second curve 332 illustrates the piecewise linear amplifier response.

  Referring to FIG. 10A, a differential representative analog signal is received from antenna 92 of FIG. Steps 150 and 151 are the low pass filtering required to remove high frequency noise from the representative signal followed by the high pass filtering used to remove power line noise. Optical nonlinear amplification 152 is used to amplify small amplitude signals more disproportionately than large amplitude signals and reduce amplitude imbalance between CG and IC signals.

Data Collection Following non-linear amplification 152 in FIG. 10A, analog information is sent to threshold comparison 156 and digitization 162 using crosspoint switch 154. If the representative signal exceeds a predetermined amplitude value, a “threshold crossing time” is established using the GPS device 130 of FIG. 6 at step 158 and placed in the time tag FIFO 132 by step 160. Simultaneous with threshold comparison 156, 12-bit resolution continuous digitization 162 at a rate of 20 MHz occurs in ADC 114 of FIG. In the LF / VLF sensor, every four samples are added together using the digital adder 134 of the FPGA 116 at step 166. As a result, a 5 MHz data sample stream having 14-bit resolution is stored in the signal FIFO of FPGA 116 by step 168. The process of moving through the FIFO 136 until the sample reaches the end is indicated by step 170. The control logic provided by the FPGA 116 and DMA controller 140 controls the flow of digital samples from the signal FIFO 136 to the ring buffer of the SD-RAM 120, which is indicated by step 172 in FIG. 10B. Inherent in the address of the data sample in the ring buffer is a time stamp to a rate value of 1 second. A separate clock signal at DSP 118 is used to coordinate the reading of data samples from ring buffer 138 into DSP CPU 144 at step 174. In practice, the transmission of data into the ring buffer in step 172 is always several samples ahead (in time) of the transmission of data out of the ring buffer and into the DSP CPU in step 174. This time lag is generally graphically represented by the “future buffer” 173 in FIG. 10B, which causes the DSP to examine several samples on either side of the sample currently being examined in step 176. Symbolize the fact that makes possible. Step 178 provides the next time stamp from the time tag buffer 142 to the CPU 144. Step 176 is a determination of whether the current data sample occurred at or after the threshold crossing time indicated by the timestamp. If so, step 180 is used to determine if the amplitude of the digitized signal was below an established threshold for the end of the event. If so, the event that caused the start of the pulse with the time stamp reproduced by step 178 is considered to be finished, ie the pulse is considered finished. Step 182 is used to advance the time tag buffer to the next event of interest and proceeds to pulse classification in step 184. If the result of step 180 is that the pulse is not considered complete and we have just started the pulse, steps 186 and 188 are the first in the current pulse with an absolute magnitude above the noise level. Used to find the time of one data sample. This time is not the same as the time stamp reproduced in step 178 because the time stamp is recorded only after the current pulse exceeds a predetermined threshold magnitude (set higher than the noise floor). Absent.

  Steps 189 and 190 are used to generate a representative electromagnetic field signal by numerically adding the current data value to the weighted sum of all previous data values, said weighting being the age of the sample Is a function of This process constitutes a digital integrator and regenerates the electromagnetic field signal in the band of interest using the differentiated electromagnetic field signal passing through this point. FIG. 11 shows the time domain response 314 of a digital integrator according to the present invention. FIG. 12 shows the frequency domain response 316 of a digital integrator according to the present invention. At the completion of step 192 of FIG. 10C, the data sample now has three components, a timestamp, the value of the derivative electromagnetic field signal, and the value of the representative electromagnetic field signal itself. The representation signal of the entire electromagnetic field pulse 293 after digital integration is illustrated in FIG.

Peak and Trough and Zero Crossing Determination Steps 194 and 196 are used to look for peaks and troughs in the representative electromagnetic field signal by searching the derivative signal for the zero crossing. In practice, the differential signal contains some noise and retains significant peaks and flags without flagging all small high frequency sign changes in the time derivative as significant peaks and troughs in the electromagnetic field signal, Smoothing 193 (digital filter processing) needs to be performed. In the preferred embodiment, this smoothing includes several samples on either side of the current sample. A first peak 294, a first trough 295, a second peak 296, and a zero crossing point 297 of an exemplary electromagnetic field signal 293 are shown in FIG. The number of peaks in the current pulse is recorded in step 200. According to step 202, if the representative signal of the electromagnetic field changes in polarity, the zero crossing time is recorded in step 204 and the counter is incremented in step 206 of FIG. 10D. Once a data sample has been tested for peak, trough and zero crossing times, the next data sample is read into the DSP's CPU 144 at step 208 in FIG. 10D and the process returns to step 174 in FIG. 10B.

  The process steps illustrated by FIGS. 10A, 10B, 10C and 10D are used to detect signals of interest and collect information. In particular, at the end of step 208, the following information has been determined: (1) the number of peaks in the current pulse of the representative signal of the electromagnetic field, (2) the number of troughs in the current pulse, (3) The time and amplitude of each peak and trough of the current pulse, (4) the number and time of zero crossings in the current pulse.

Pulse Classification If it is determined that the pulse has ended in step 180 of FIG. 10B, the process proceeds from step 184 to step 242 of FIG. 10E, where pulse classification begins. Pulse classification begins by establishing a default value for the pulse as an IC event in step 244. The total pulse duration is evaluated at step 246. If the pulse is too short in duration or the maximum amplitude of any peak in the pulse is too small, the event is evaluated as noise and discarded in step 248 and zero crossings, peaks and trough counts are zeroed in step 250. The process then returns to step 174 of FIG. 10B. If the pulse has sufficient duration, it is evaluated at step 252 for an excessive number of zero crossings. Too many zero crossings is another indicator of noise, in which case the pulse is discarded in step 248 and the process returns to step 174 in FIG. 10B. The new pulse classification process is illustrated by step 256, where the pulses are evaluated in a short time from the maximum peak of the first polarity to the maximum peak of the opposite polarity. The result of this test must be temporarily saved (steps 257 and 258) since after this test all pulses proceed to the bipolar amplitude ratio test in step 260 of FIG. 10F. This pulse classification parameter is given by the time between the first peak 294 and the second peak 296. Pulses with a short time difference between these peaks are classified as IC pulses or leader pulses.

The bipolar amplitude ratio test step 260 determines whether the maximum of any subsequent opposite polarity peak (299 in FIG. 13) is greater than a predetermined percentage value of the maximum peak 294 of the first polarity. used. In events where the opposite polarity peak time difference (256 in FIG. 10E) is short, the bipolar amplitude ratio distinguishes between bipolar IC discharge pulses and leader pulses that tend to be close to unipolar. Therefore, after step 260, the results of the opposite polarity peak time test must be examined (steps 261, 263). If the bipolar amplitude ratio 260 is large and the opposite polarity peak time result is true (step 263), the pulse is classified according to the reader (step 264). In the preferred embodiment, the leader pulse is considered a special case of an IC discharge pulse. Beyond steps 264 and 265, no other tests are required and the IC discharge pulse information is saved in step 284 of FIG. 10H (267). On the other hand, if the bipolar amplitude ratio is large, the opposite polarity peak time result is false (step 263) and the pulse is classified in step 270 as due to a distant CG lightning strike. Eventually, if the bipolar amplitude ratio is small and the opposite polarity peak time result is false (step 261), the pulse is classified in step 268 as CG return lightning strike (not far). In any event, pulse classification continues at step 271. Optionally, if classification is used to remove the IC discharge, the pulse may be discarded at step 266 and the process returns to step 174 of FIG. 10B.

Peak-to-zero test The first peak for zero time is evaluated in step 271 of FIG. 10F. If the peak-to-zero time 298 in FIG. 13 is relatively long in duration, the pulse retains its classification as a CG event. If the peak-to-zero time 298 is short, the pulse is reclassified as an IC event in step 262. In either event, pulse classification continues at step 274 of FIG. 10G.

Second Peak Test Greater Than First Peak Step 274 in FIG. 10G determines whether two or more peaks of the same polarity (294, 296 in FIG. 13) occur before peaks 299 of opposite polarity. If not, the pulse classification ends at step 282. Otherwise, an additional test 276 is used to determine whether the second peak 296 of the same polarity in the pulse is sufficiently larger than the first peak 294 of the pulse. If true, the pulse is reclassified as a cloud discharge in step 280. Otherwise, the pulse keeps its classification as a CG discharge. In either case, the classification ends at step 282 and the process proceeds to record pulse information, step 284 of FIG. 10H.

Save Pulse Information Step 284 of FIG. 10H begins the process of storing pulse information and event classification. In step 286, the pulse is defined by the following information: threshold crossing time, start time, first and second peak times, first rise time, first and second peak amplitudes, peak to two. Zero time, reverse polarity peak time and amplitude, electromagnetic field derivative at the first zero crossing, pulse classification. In step 288, this information is stored in the SD-RAM result buffer 146 in FIG. At this point, the process obtains a new data sample (step 289) and returns to step 174 of FIG. 10B. At the same time, the pulse information stored in the result buffer at 288 is made available to other processes 290 that find and process the pulse train.

  The flowchart of FIG. 14 illustrates the process of grouping individual pulses into pulse trains. Pulse train processing begins at step 412 and begins when pulse information is stored in result buffer 146 at step 288 of FIG. 10H. The time between the most recent pulse and the previous pulse is evaluated at step 414. If the period between two pulses exceeds a predetermined amount, the previous pulse is examined at step 416 to determine if it is part of an existing pulse train. If so, the existing pulse train is considered complete, the data is compressed in step 422, and the compressed data is sent to the host PC 122 in step 424, where the previous pulse and its associated pulse train are The result buffer 146 is removed at step 426 and the pulse train buffer is cleaned at step 428. If the previous pulse is not part of an existing pulse train, the intervening pulses are transferred to the host PC 122 at step 418 along with all parameters stored at step 286 of FIG. 10H and removed from the result buffer 146 at step 420. Is done. If the time period evaluated in step 414 is found to be less than the predetermined amount, in step 430 the previous pulse is evaluated to determine if it belongs to an existing pulse train. If so, an optional PTK test is performed at step 432 and at step 436 the amplitude of the latest pulse is evaluated to determine if it is greater than any previous pulse in the current pulse train. If it is found that it is greater than any previous pulse in the current pulse train, the latest pulse is flagged at step 438. Returning to step 430, if the previous pulse is not part of the current pulse train, then in step 434, a new pulse train starting with the previous pulse is formed and the process proceeds to steps 432, 436 and 438. The final step in this process is to move the latest pulse to the position of the second latest pulse in the result buffer 146 in step 440 and then return to step 412 to wait for the arrival of the next pulse. The maximum latency is also used to ensure that pulses are transmitted to the central analyzer within a reasonable period of time when there is little activity.

  In previous generations of lightning detection systems, the “pretrigger suppression” test consists of determining whether a CG lightning strike is immediately preceded by an event of opposite polarity. If so, both the current pulse and the previous pulse are discarded as they are probably caused by IC discharge or noise or interference. In these conventional systems, discarding these pulses was considered more efficient than constraining the analog hardware to further analysis. However, it has been found that it is more efficient to perform a more complete pretrigger suppression test in the present invention. Referring to FIG. 15, the flowchart illustrates the pre-trigger suppression test referenced in FIG. This test begins at step 450. The current pulse is evaluated at step 452 to determine if it is a CG lightning strike. If not, the process returns to pulse train processing algorithm step 466. However, if the current pulse is a CG lightning strike, it is first transmitted to the host PC in step 453, and the complete feature data set is saved in step 286 of FIG. 10H. The previous pulse is then examined at step 454 to determine if it is an IC event. Otherwise, the process returns to the pulse train processing algorithm in step 466. If the current pulse is the result of a CG lightning strike and the previous pulse was caused by an IC event, the next step 456 of the algorithm is to determine if the current pulse and the previous pulse have opposite polarities. is there. Otherwise, the process returns to the pulse train processing algorithm in step 466. If both pulses have opposite polarities, the current pulse train is discarded at steps 458-464 and the process returns to step 412 of FIG.

  As mentioned in the summary of the present invention, the RPS sensor can detect and evaluate at least one digit more lightning discharge than the previous generation sensor. Therefore, an improved method for sending lightning information to a central analyzer is needed. The present invention uses a new data compression process for lightning discharge produced in the pulse train by the process of FIG. The maximum pulse is designed because the reference pulse and its amplitude, time and direction (if available) are included in the 88-bit data record. Other pulses in the pulse train are represented by an amplitude that is a percentage value of the reference pulse and a time stamp relative to the time of either the previous or next pulse, and represented by a 24-bit data record. This greatly reduces the information that must be transmitted.

  Referring to FIG. 16, six electromagnetic pulses 470 are listed with their related fractional second time 472, time 474 in microseconds, and amplitude 476. These pulse examples occurred at midnight on August 6, 2001. The first pulse occurs 500 milliseconds after midnight on this day and is represented in detail in FIG. The complete time stamp of the maximum pulse in this example is from January 1, 1970 (UNIX time) to 97056000 seconds. The 32-bit hexadecimal representation 480 of this time stamp is 3B6DDE00h. The time count of the percentage value is derived by dividing 500 microseconds by a predetermined time unit, which is 50 nanoseconds in a preferred embodiment (each 50 nanoseconds) for the use of a 20 MHz ADC. Second cycle is "50 nanosecond count"). 500 microseconds divided by 50 nanoseconds results in a time count of 10,000 rate values. The hexadecimal representation 482 of the time count of this percentage value is 2710h. The hexadecimal representation 484 of the amplitude (4400 counts) of the first pulse 471 is 1130h. This particular sensor gives no direction information and indicates by specifying a value of 359.999 degrees as the angle measurement. The hexadecimal representation 486 of this angle information is 7FFFh. The determination that this pulse was generated by an IC discharge is indicated by a binary value 488 of “1” (CG lightning strikes generate “0” bits).

Data Compression The illustrations of the flowcharts of FIGS. 18 and 19 and FIGS. 20A-20D are used to illustrate the process of compressing a pulse train during data recording.

  During the grouping of pulses (FIG. 14), the maximum amplitude pulse is confirmed. In this example, the second pulse 473 of FIG. 16 is a maximum with an amplitude of 5100 counts. The binary representation of the maximum amplitude pulse 473, including time, fractional value of second count, amplitude, angle, classification, and an additional bit that indicates whether the amplitude that results in the fractional value of the smaller pulse is included, is Generated by the present invention. FIG. 18 shows an 88-bit compressed representation 490 of the maximum pulse for this example according to the compression algorithm. The hexadecimal representation of the maximum amplitude pulse 473 is shown by the binary representation 490 in FIG.

  According to the present invention, pulses in the same pulse train other than the maximum amplitude pulse 473 are represented by an amplitude that is a ratio value to the maximum pulse 473 and a time difference with respect to the previous or next pulse. For each pulse before the maximum pulse, the time difference is established by subtracting the 50 ns time count of the current pulse from the 50 ns time count of the following pulse, resulting in a 13-bit signed binary that is negative. Yield number. Similarly, for each pulse after the maximum pulse, the time difference is given by subtracting the 50 nanosecond count of the current pulse from the 50 nanosecond count of the previous pulse, resulting in a positive 13-bit signed value. Produce. For all pulses other than the maximum pulse, the amplitude resulting in the pulse percentage is calculated by multiplying the current pulse amplitude by 1000, dividing by the maximum pulse amplitude, and rounding the result to the nearest whole number. This integer is then represented by a 10-bit binary number. The 1-bit binary flag is used to indicate whether the pulse is due to a CG event (“0”) or an IC event (“1”). The 13-bit time difference, the 1-bit classification flag, and the 10-bit percentage amplitude are concatenated together to produce the binary representation 500 in FIG. 19 and their hexadecimal equivalent 502. Thus, the six pulses of FIG. 16 are represented by one 88-bit data record 490 of FIG. 18 for the maximum amplitude pulse and five 24-bit data records 500 of FIG. 19 for five pulses other than the maximum pulse. Is done. Other optional embodiments of the present invention use a 15-bit signed time difference value and an 8-bit ratio value amplitude value than the 13-bit time difference and 10-bit ratio value amplitude referred to above. Other versions of this 23-bit data recording design may be used to account for the trade-off between maximum duration and amplitude resolution.

  20A-20D are used to illustrate the compression process according to the present invention. In step 504, the position of maximum amplitude is recognized. In step 506, the 32-bit timestamp of the maximum pulse and the 25-bit 50 nanosecond count are stored. If the amplitude percentage value is to be used in the representation of other pulses in the pulse train (step 508), the amplitude percentage value flag bit is set in step 512, otherwise it is cleared in step 510. A 14-bit binary representation of the maximum pulse amplitude is stored in step 514. If direction information is provided (step 516), a 15-bit binary representation of the maximum pulse azimuth is generated in step 520, otherwise a 15-bit binary representation for 359.999 degrees is used in step 518. . The process continues at step 522 of FIG. 20B where a flag indicating the type of event is set. Binary “1” indicates an IC event, and “0” indicates a CG event. The “current pulse” pointer is set in step 524 to a data structure representing the pulse before the maximum pulse. A 13-bit value is generated in steps 528 and 530 for the time difference between the current pulse and the immediately following pulse in the pulse train. In step 532, the process determines whether an amplitude percentage value is used, and if so, the 10-bit value is determined in steps 534 and 536 with respect to the amplitude percentage value of the pulse immediately before the maximum pulse. The classification bit is generated and set in step 538 of FIG. 20C. The pointer initialized in step 524 is then decremented in step 540 and the process loops to step 526 in FIG. 20B until the pointer reaches the start of the pulse train.

  When the pointer reaches the beginning of the column, it indicates that all pulses before the maximum pulse have been encoded, and the pointer is reset to the first pulse after the maximum pulse (step 542 in FIG. 20C). A 13-bit value is generated in steps 546 and 548 of FIG. 20C for the time difference between the maximum pulse and the previous pulse in the pulse train. In step 550 of FIG. 20D, the process determines whether an amplitude percentage value is being used, and if so, the 10-bit value is the amplitude percentage value of the pulse immediately following the maximum pulse in steps 552 and 554. The classification bit is set in step 556 of FIG. 20D. The pointer set in step 542 is incremented in step 558 and the process loops to step 544 in FIG. 20C until the pointer reaches the end of the pulse train. If this occurs, the information is compressed according to the present invention and is ready to be transmitted to the central analyzer in step 560 of FIG. 20D, at which point the process ends (step 562).

  Transmission of information between the remote programmable sensor 12 and the central analyzer 20 occurs as a packet or message. These messages are made up of several 8-bit binary words. The size of the packet is variable depending on the number of pulses in the pulse train. The more pulses in the pulse train, the larger the message. Each message includes 11 bytes representing the maximum pulse in the pulse train. The message also includes the other 3 bytes for each additional pulse in the pulse train if the amplitude percentage value is used. Otherwise, the message contains 1 and 3/4 (actually 2) bytes for each additional pulse in the pulse train.

  Sometimes, even in sensors operating in the VHF frequency range, even pulse train compression is insufficient to allow all pulses to be transmitted by available communication systems in real time. In such a situation, the sensor can only select and transmit only some of the pulses or pulse trains that occur in a particular time period. More particularly, the time is divided into short periods (eg, 50 millisecond periods) relative to the start of the second. In the event of information backup in a communication system (as evidenced by data accumulation in the output buffer), the amount of data decreases to the first few events in the set of these periods. For example, we can only transmit the first event from each of the 20 50 millisecond periods in 1 second. This “data decimation” algorithm is synchronized between all sensors in the network by GPS time information, so that all sensors are guaranteed to transmit data corresponding to the same period. The operation of the central analyzer is described below.

Data decompression When information is received by the central analyzer, it must be decompressed. The flowchart of FIGS. 21A-21C represents the process necessary to decompress a message received from a remote programmable sensor. The number of bytes in the message is determined in step 572 of FIG. 21A. The maximum pulse time stamp is extracted from the message in step 574 and the use of the amplitude percentage value is determined in step 578. The total number of pulses in the message is determined in step 580 if the amplitude percentage value is not used, and in step 582 if the amplitude percentage value is used. This algorithm takes the number of bytes in the entire message determined in step 572 and subtracts 7 words, the number of words needed to describe the maximum amplitude pulse. This result is divided by the number of bytes required to represent the compressed pulse, 2 if the amplitude percentage value is not used, and 3 if the amplitude percentage value is used. A quantity of 1 is added to the quotient to generate the total number of pulses in the pulse train.

  The maximum pulse amplitude is determined in step 584 and the event classification is determined in step 586. The loop counter is initialized to zero in step 588 of FIG. 21B. The loop is used to proceed through all remaining pulses in the pulse train. Each signed time difference is extracted from the compressed data stream in step 592 and determined in step 602 or decision step 596 if it is the first pulse through the loop (as determined in decision step 594). As such, if the time difference changes sign, it is added to the time stamp of either step 600) or the time of the most recently decoded pulse (step 598) to generate the absolute time that each pulse occurred. To do. The pulse classification is determined in step 604. Step 606 of FIG. 21C is used to determine whether an amplitude percentage value is included in the compressed data. If so, the actual amplitude of each pulse is recovered in steps 608 and 610 of FIG. 21C. The loop counter is incremented at 612 and the process returns to step 590 of FIG. 21B. The loop continues until the loop counter is equal to the number of pulses determined in step 580 or step 582, indicating that all pulses in the pulse train have been stretched. At this point, the process ends at step 614 of FIG. 21C.

  Once the information is decompressed, it is compared with information from other remote programmable sensors 12 to determine the location, amplitude and time of occurrence of the lightning event. With the increased information provided by detecting the pulse train generated by the same lightning event, improved correlation techniques are needed.

Pulse Correlation Four pulse trains are illustrated in FIG. The first pulse train 650 is detected by the remote programmable sensor 12 closest to the location of the lightning activity, as indicated by the earliest arrival time from the lightning event to the sensor and the high pulse number (13). The second pulse train 652 is a remotely programmable sensor located farther from the lightning activity than the sensor that generated the first pulse train 650, as indicated by later arrivals and fewer (8) pulses of smaller amplitude. 12 is detected. The remote programmable sensor 12 for the third pulse train 654 is located further from the lightning activity than the sensor generating the second pulse train. The third pulse train 654 has only four pulses whose amplitude is significantly smaller than that of the first or second pulse train. Fourth pulse train 656 is detected by remote programmable sensor 12 furthest from lightning activity. Only three pulses are evident in this pulse train.

  The unique correlation approach according to the present invention uses either three remote programmable sensors 12 or two remote programmable sensors 12 with direction information and provides a reliable pulse source position. If direction information is not provided by the remote programmable sensor 12, a theoretical ideal subset of three sensors is selected to provide an initial TOA based position. This logical set of sensors is used to reduce other possible correlated pulse sets. To determine the required TOA difference, all pulses in the time window for each sensor are sorted by reducing the signal amplitude generating the table of FIG. The three sensors with the largest pulses at the top of these amplitude sorted lists are then selected as the ideal subset described above. This selection is based on the fact that the maximum amplitude pulse is generated by a substantially vertical lightning discharge, and the maximum pulse detected in one sensor is likely to be the maximum pulse detected in another sensor. In the example of FIG. 22, the subset includes a remote programmable sensor 12 that generates a first pulse train 650, a second pulse train 652, and a third pulse train 654. These pulse trains are selected from the pulse trains reported by all sensors because they possess the largest amplitude pulses. The maximum pulses from each of the three ideal sensors are 1H, 2F and 3C as shown in FIG. Using the time between these three pulses, the estimated TOA based location and generation time is determined for the discharge. Next, the expected value of travel time from the estimated discharge position to all sensors is determined and the pulse train is adjusted in time by subtracting the relative time differences alpha, beta and gamma shown in FIG. Used for. The alpha, beta, and gamma values used to slide the second pulse train 652, the third pulse train 654, and the fourth pulse train 656, respectively, are calculated from the time it takes for the signal to reach each of the associated sensors. Given by the difference minus the time it takes to reach. After adjustment, the arrival times of the maximum amplitude pulses in the first three sensors coincide with pulse 4B of the fourth pulse train 656, as shown by the grouping of pulses 660 in FIG. Similarly, the second largest correlated pulse coincides with 4C of the fourth pulse train 656, as shown by grouping 662 in FIG. 24, and finally the third largest pulse is as shown by grouping 658. , Corresponding to 4A of the fourth pulse train 656. If it is found that at least one, preferably more inter-pulse periods (as in FIG. 24) are matched, the probability of having the correct correlation between the various sensors is close to one. If the pulse train is not known to match as in FIG. 24, different combinations of pulses from the three sensors from the amplitude sorted list in FIG. 23 are used to generate a new estimated discharge position. And the process begins to finish. This process, in the preferred embodiment, is only attempted 3 or 4 times, but may be repeated any number of times until the pulse is exhausted. This correlation process has the advantage that it works well in the presence of regularly spaced pulses of similar amplitude.

  An alternative correlation process is used for a lightning detection system having at least two sensors with direction finding capabilities. The correlation process is described above, except that angle and time measurements from just two sensors are used to generate an initial assessment of discharge location and time of occurrence, from which travel times to other sensors are evaluated. The method is the same.

  Yet another extension of the correlation process is used to track the travel path of lightning discharge propagation that has a fundamental horizontal component for these movements. Recognizing correlated pulses from different sensors with slightly different inter-pulse periods indicates an expansion or shortening of the pulse train caused by the horizontal movement of lightning during the generation of the pulse train. Either of the first two correlation methods may be used as long as a large degree of timing is allowed in fine scale pulse correlation to account for slightly different inter-pulse spacing. If two or three sources can be positioned, the propagation speed and direction may be evaluated. These estimates may then be used to predict pulse train shortening or expansion from other sensors to provide further improvements to the correlation process. This correlation technique is useful for separating different branches of lightning discharges. The position based on the TOA is determined through the process described in the previous paragraph, not only to obtain the final optimized position when there are more than 4 TOAs, but also to obtain the initial position evaluation (starting position).

  In view of the above, it can be seen that the system according to the present invention provides a number of advantages over prior art lightning detection and analysis systems. First, the system has the ability to detect both CG and IC events while evaluating dead time. Second, remote programmable sensors have increased sensitivity to IC events without being overwhelmed by CG events by analysis of differentiated and / or compressed waveforms. Third, the system has the ability to transmit increased event information to a central analyzer over a limited broadband communication channel using a new compression and decompression scheme. In addition, the system has the ability to use new correlation techniques to correlate events that produce complex pulse trains in the central analyzer, thereby providing an accurate TOA based position.

  The terms and expressions used in the above specification are used herein as descriptive terms, not as limiting terms, such terms and features excluding features shown and described or some equivalent thereof. It is recognized that there is no intention in the use of the expression and that the scope of the invention is defined and limited only by the claims.

FIG. 4 is an illustration of the physical layout of an exemplary system for obtaining data for cloud-to-ground and in-cloud discharges according to the present invention. FIG. 4 is an illustration of a representative LF / VLF electromagnetic field waveform and corresponding time derivative waveform generated by a cloud-to-ground discharge. FIG. 4 is an illustration of a representative LF / VLF electromagnetic field waveform and corresponding time derivative waveform generated by an IC discharge. Fig. 4 shows an empirically obtained cumulative distribution of signal amplitude distributions normalized for cloud-to-ground discharge and IC discharge ranges. 1 is a functional block diagram of a preferred embodiment of a lightning detection system and data acquisition system according to the present invention. FIG. 1 is a structural block diagram of a preferred embodiment of a lightning detection system and data acquisition system according to the present invention. FIG. 1 is a block diagram of an antenna filter network according to the present invention. FIG. 8 is a graph of two exemplary frequency responses of the antenna filter network of FIG. It is a graph of the gain of two nonlinear amplifiers. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. 6 is a flowchart of the operation of a preferred embodiment of a detection and data acquisition method for individual lightning discharge pulses according to the present invention. Fig. 4 shows the time domain response of a digital filter according to the invention with respect to the numerical integration of the signal representing the time derivative of the electromagnetic field. Fig. 3 shows the frequency domain response of a digital filter according to the invention with respect to the numerical integration of the signal representing the time derivative of the electromagnetic field. 2 illustrates an analysis performed on a general electromagnetic field waveform by a system according to the present invention. 4 is a flowchart illustrating a process for grouping individual pulses into a pulse train according to the present invention. 6 is a flowchart illustrating a “pretrigger suppression” test. It is a table | surface which shows the time and amplitude of six electromagnetic pulses which form a pulse train cumulatively. It is a table | surface which shows the decimal system and hexadecimal system expression of the 1st pulse of FIG. 16 illustrates binary and hexadecimal representations of the maximum pulse of FIG. 16 compressed according to the present invention. 17 illustrates binary and hexadecimal representations of all pulses other than the maximum pulse in the pulse train of FIG. 4 is a flowchart illustrating a compression process according to the present invention. 4 is a flowchart illustrating a compression process according to the present invention. 4 is a flowchart illustrating a compression process according to the present invention. 4 is a flowchart illustrating a compression process according to the present invention. 4 is a flowchart illustrating an expansion process according to the present invention. 4 is a flowchart illustrating an expansion process according to the present invention. 4 is a flowchart illustrating an expansion process according to the present invention. 22 is a time graph illustrating the arrival of electromagnetic pulses at different remote programmable sensors as regenerated by the stretching process of FIGS. 21A-21C. FIG. 23 is a table for the pulses of FIG. 22 sorted by amplitude and sensor position according to the present invention. 4 is a timed graph illustrating a correlation process according to the present invention.

Claims (6)

A lightning detection system including a lightning detection sensor,
The lightning detection sensor
A sensor unit for detecting electromagnetic fields generated by lightning,
A digital processor for generating digital data representing the detected electromagnetic field;
The digital data is
Defining at least one pulse having one or more peaks of a first polarity and one or more peaks of a second polarity;
The digital processor is
Identifying a maximum value of one or more peaks of the first polarity, a maximum value of one or more peaks of the second polarity, and a time between the maximum values with reference to a predetermined clock. And comparing the time with a predetermined reference value, and classifying the pulse as being related to IC discharge when the time is less than the reference value.   The line sub is adapted to generate an analog signal representative of a time derivative of the electromagnetic field, and the lightning detection sensor comprises an analog-to-digital converter for digitizing the analog signal, The lightning detection system of claim 1, wherein the processor is adapted to integrate the digitized analog signal to generate the digital data.   The system of claim 2, wherein the digital processor is adapted to identify the peak by determining a zero crossing of the digitized analog signal. A lightning detection method,
The lightning detection sensor
Detecting electromagnetic fields generated by lightning,
Digital data generation step for generating digital data representative of the detected electromagnetic field, wherein the digital data has at least one pulse having one or more peaks of a first polarity and one or more peaks of a second polarity; A digital data generation step,
Identifying a maximum value of one or more peaks of the first polarity, a maximum value of one or more peaks of the second polarity, and a time between the maximum values with reference to a predetermined clock. Steps,
Comparing the time with a predetermined reference value;
Classifying the pulse as relating to IC discharge if the time is less than the reference value;
A method of detecting lightning, comprising: Generating an analog signal representing the time derivative of the electromagnetic field;
Digitizing the analog signal;
Integrating the digitized analog signal to generate the digital data;
The lightning detection method according to claim 4, further comprising:   6. The lightning detection method according to claim 5, further comprising the step of identifying the peak by determining a zero crossing of the digitized analog signal.

Images