• Main
• System Description
• Station Locations
• Publications
• Contacts
• Acknowledgments
• Links

• Real Time Data

• Real Time
• Hourly Images
• Daily Images
• Monthly Images

• Reprocessed Data

• 2001-2004 Data Set
• 15 minutes
• Daily Images
• Monthly Images
• FTP

• AMMA Data Set

• 15 minutes
• Daily Images
• FTP

Zeus long-range lightning detection system is based on detection of sferics - the impulsive radio noise emitted by a lightning strike - in the Very Low Frequency (VLF) spectrum between 7 and 15 kHz. In the VLF portion of the radio spectrum sferics propagate thousands of kilometers through the earth-ionosphere waveguide. Zeus consists of a network of six VLF receivers located around the periphery of Europe. Each receiver reports the vertical electric field as a time series which represents the sferic's waveform, and includes a time stamp synchronized by the Global Positioning System (GPS) clock. The Arrival Time Difference (ATD) between the time series from pairs of receivers is extracted by time correlation. Each ATD yields an elliptic locus-of-points on the earth's surface for the lightning location, and the intersection of several ellipses from a sferics candidate defines a ''fix'' location. The six receivers are located in Birmingham (UK), Roskilde (Denmark), Iasi (Romania), Rhodes (Greece), Mt. Etna (Italy), and Evora (Portugal). Resolution Displays´ technology has evolved from development under NASA SBIR contract NAS5-32825 and earlier work done by Dr. Tony Lee at the UK Met Office. The following sections present a review of system hardware and software.

1. System Hardware

The system hardware can be separated in two parts: a number of receivers and a central computer station. A simplified sketch of this system is illustrated in Figure 1. Each receiver consists of an outside VLF antenna, preamplifier, GPS timing generator and signal converter, and an inside personal computer (PC) with Internet communication access. The central computer station runs on a PC, which receives via Internet the electric field measurements from each receiver. A more detailed diagram of the receiver antenna's components is shown in Figure 2. Enclosed in an small electrical utility box mounted on a two meter high pole, the antenna is located outside the building some 30 meters or more to avoid electrical noise. The VLF electric field signal is pre-amplified and encoded by an analog-to-digital converter while an integral GPS receiver provides timing information. The digitized data are then sent inside to a PC which executes a digital signal processing algorithm to remove noise, the identification algorithm that detects probable sferics candidates, and data compression algorithms to compress files sent to the central station over the Internet. The VLF receiver hardware has a dynamic range of 100 dB and timing accuracy within one microsecond of GPS time. The receiver's noise floor is 100 nanoVolts/meter/root-Hertz.

Figure 1. Overview of network architecture.

Figure 2. Sketch of antenna receiver’s component.

2. Software Overview

The system software is divided between the receivers and the central station. The receiver signal processing algorithms are optimized to separate distant, and therefore weak, sferics from the interference that surrounds them. Signal quality control is integrated across the system to eliminate low-quality sferics data that could cause false location reports. The receiver software is capable of capturing 130 sferics-per-second. Receiver bandwidth is defined by a finite-impulse-filter (FIR) digital filter extending 4.1 kHz above and below its center at 11.0 kHz. Each sferic waveform is contained in a 4.4 millisecond window. The wave-shape information is heavily compressed to about 160 bits per sferic. This compressed sferic data are accumulated into files of 16 seconds duration, which have a typical size of 5 to 20 kilobytes. These files are backed-up and transmitted to the central station. Once the data reaches the central station two tasks are performed: (a) decompression and correlation, and (b) locating and optimization. In the decompression task each file is uncompressed and the waveform signal is restored. It follows that same-source candidates observed by the different receivers are compared to extract the corresponding ATD values. Namely, the waveform signals from two receivers are analyzed and the time difference of the highest cross-correlation value defines the ATD. Accordingly, ATD values are computed for all possible combinations of receiver pairs. In the present system with six receivers fifteen ATD values are computed. As mentioned earlier, these ATD values represent elliptic positions between two receivers with same signal arrival time difference. The intersection among those ellipses defines a sferic fix.

In the second task an ATD technique serves as the primary locating algorithm. All the ATDs for a same-source candidate are examined to estimate the strike location and time using a least-squares fit weighted selectively to certain ATDs. The algorithm adaptively compensates for correlation ambiguities, a problem that increases with long propagation distances. This is a consequence of the distortion of the electrical signal traveling the differing propagation paths to the receivers and possible different sferics noise observed at the same time. These effects might contribute to distort the waveform and therefore create several ambiguous peaks. A final optimization takes into account the likely errors associated with each ATD in light of the estimated location and is dependent on the network's geometric configuration. Proper resolution of these ambiguities is an important task as it increases the system accuracy and sensitivity especially at long distances.