accurately capture lightning flashes within the frequency range of 1Hz to 10MHz. The lightning detection system

was fully assembled using a Fast Field buffer circuit integrated with a parallel-plate antenna to ensure accurate

detection of lightning flashes. The system successfully detected a total of 60 negative cloud-to-ground (-CG)

flashes and 51 intra-cloud (IC) flashes on 11 July 2023 between 01:32:00 and 08:3:00. Analysis of the –CG

flashes indicates that the majority of 54 flashes consisted of single return strokes, while only 6 flashes exhibited

subsequent return strokes which 4 flashes with two subsequent return strokes and 2 with three subsequent return

strokes, respectively. The developed system achieved an 83% consistency in identifying lightning flashes when

comparing with an existing lightning detection system, demonstrating its potential as a low-cost yet reliable

dislodge electrons from atoms in the surrounding air, causing a transient and high-current electric discharge. The

path length of lightning discharges is typically measured in kilometers. Another definition of lightning is an

electrical discharge taking the form of a spark within a cloud that is charged (Kostinskiy et al., 2016). A typical

lightning strike is composed of three to four strokes, although it can involve more. Each re-strike is separated by

Lightning can be classified into various types which are cloud-to-cloud (CC), intra-cloud (IC), cloud-to-ground

(CG). When focusing on the CG, further classification can be based on polarity, resulting in downward negative

lightning, upward negative lightning, downward positive lightning, and upward positive lightning (Kalair, 2013;

Shivalli, 2016). It can also be divided into positive (+CG) and negative (-CG) cloud-to-ground based on the

polarity. Many studies have been conducted in various geographic regions found that around 10% of global CG

The cost of existing lightning detection systems is quite expensive. The existing system uses high-cost

components, and specialized sensors that can detect electromagnetic pulses associated with lightning (Adzhiev,

2013; Bitzer, 2015). Although these advanced systems provide accurate tracking and location capabilities, but

the cost were still expensive. This creates a demand for the development of alternative solutions that offer

comparable accuracy and reliability at a more affordable price point. By addressing cost issues and proposing

low-cost lightning detection systems, lightning detection becomes more accessible to a wider variety of users

and industries. Such a system would use cost-effective components and technology and ensure accurate lightning

detection. To overcome the high cost of existing lightning detection systems and monitor lightning events, the

The process of designing the lightning detection system involved assembling several components such as

antenna, coaxial cable, Fast Field buffer circuit, Pico Scope, and laptop as shown in Figure 1. A parallel plate

antenna constructed using metallic material have been used for lightning detection. The antenna used in this

design had a capacitance of 59pF. The designed antenna was linked to the buffer circuit through a short 60cm

coaxial cable with a 50Ω impedance to establish the connectivity. The capacitance of the coaxial cable is 60pF.

Subsequently, the buffer circuit was connected to the Pico Scope via a longer coaxial cable, and the Pico Scope

was interfaced with the laptop. Upon triggering by lightning flashes, the system collected and stored all the data

The installation setup of a comprehensive lightning detection system was done in real time measurement at the

rooftop UTeM building. This emphasizing key considerations such as accessibility, unobstructed exposure to the

open sky and the absence of significant barriers that might impede detection accuracy. The setup entails the

meticulous installation of a lightning detection system comprising a parallel plate antenna, a fast field buffer

circuit, and a coaxial cable. Each component was methodically positioned and configured to ensure optimal

functionality and accurate lightning detection. Strategic placement of lightning detection system across the

rooftop was executed to maximize coverage area. Many factors such as the field of view, elevation angles, and

This section will present the results and discussion obtained from the development of lightning detection system.

The signal received will be process and analyse throughout the observation. Then, the results will be compared

The results obtained shows a total of 111 lightning flashes were detected using the low-cost lightning detection

system. These lightning flashes were detected within a small radius of topical thunderstorms in Malacca,

Malaysia (UTeM). The entire dataset has been analysed and categorized into several types of lightning such as -

recorded data includes 60 -CG and 51 IC lightning flashes as shown in Figure 4. The separated data comprises

the selected data with a strong flash signal. Majority of the -CG flashes were found to have a single strokes and

others consisted of two or more return strokes. In detail, out of the 60 recorded –CG flashes, 54 were single

Figure 6 shows the characteristics of a return stroke captured on July 11, 2023, from 01:32:00 to 08:30:00. The

results indicate three characteristics of –CG flashes were recorded a single return stroke, two return strokes, and

Figure 6. The return stroke captured by the lightning system (a) -CG flashes (b) -CG flashes with two return

Figure 7-9 illustrates the comparison of return strokes captured by the lightning system FF1 and FF2 for –CG

flashes with a single return stroke, two and three return strokes. Analysis of the –CG flashes indicates that the

majority of 54 flashes consisted of single return strokes, while only 6 flashes exhibited subsequent return strokes

difference. Similar measurement applies to both systems to quantify how related or close data samples are to

each other. The similarity measure is usually expressed as a numerical value, typically between 0 and 1, where

0 represents low similarity (dissimilar data objects) and 1 represents high similarity (very similar data objects).

The analysis of 60 flashes results in a mean similarity value of 0.8266. This indicates similarity performance

achieved in the 60 flashes was around 83%. Through the analysis obtained, 0.9936 maximum similarity was

achieved while minimum similarity is 0.7149. The parameter of similarity between system FF1 and FF2 as

The authors would like to acknowledge the support provided by Faculty of Electronics and Computer

support and resources. This project also is part of the output UTeM Short Term Grant