Simulating Lightning Discharges: The Influence of Environmental Conditions on Ionization and Spark Behavior

Abstract

This study investigates the behavior of spark discharges under various environmental conditions to simulate aspects of early-stage lightning dynamics, with a focus on their spectral characteristics, propagation, and ionization behavior. In a laboratory setting, spark discharges generated by a Tesla coil operating with high-frequency alternating current (AC) were analyzed under varying air humidity and water surface conductivity. Spectral analysis revealed that the discharges are dominated by the second positive system of molecular nitrogen N₂ (2P) and also exhibit the first negative system of molecular nitrogen ions N₂⁺ (1N). Notably, the N₂ (2P) emissions show strong peaks in the 350–450 nm range, closely matching spectral features typically associated with corona and streamer discharges in natural lightning. Environmental factors significantly influenced discharge morphology: in dry air, sparks exhibited longer and more branched paths, while in moist air, the discharges were shorter and more confined. Over water surfaces, the sparks spread radially, forming star-shaped patterns. Deionized (DI) water, with low conductivity, supported wider lateral propagation, whereas higher conductivity in tap water and saltwater suppressed discharge spread. The gap between the electrode tip and the surface also affected discharge extent and brightness. These findings demonstrate that Tesla coil discharges reproduce key features of early lightning processes and offer insights into how environmental factors influence discharge development.

1. Introduction

The beauty and energy of lightning make it a fascinating topic to study. Despite being a common atmospheric event in the world, many aspects of its underlying mechanisms remain incompletely understood. Lightning is a natural atmospheric phenomenon involving high-voltage electrical discharges. Even with its frequent occurrence worldwide, many aspects of its initiation and propagation mechanisms remain incompletely understood. When charged regions within clouds induce an opposite charge on the ground, the resulting electric potential difference can exceed the breakdown threshold of air, triggering a cloud-to-ground (CG) lightning discharge [1]. Over land, lightning is often dominated by CG flashes, whereas over oceans, lightning tends to be less frequent but optically more intense [2,3,4,5]. This contrast is thought to be due in part to the higher conductivity of seawater, which facilitates charge redistribution and may enhance peak currents and light intensity [4,5,6,7]. Moreover, it has been reported that lightning becomes brighter as the salt concentration in water increases [4]. These observations suggest that environmental conductivity plays a significant role in the development and appearance of lightning discharges.

A Tesla coil generates high-voltage, high-frequency spark discharges from a pointed electrode, producing ionization and plasma arcs in air. Although natural lightning is primarily a DC phenomenon, its initiation phases, such as corona and streamer discharges, involve rapidly changing local electric fields that can be approximated by AC-like behavior [8]. Therefore, despite its high-frequency AC nature, the Tesla coil provides a valid experimental analog for studying early lightning initiation processes under controlled conditions. Figure 1a,b represent the schematic diagram of the ionization process and electron avalanche initiation, respectively. Electrons accelerated from the electrode tip collide with air molecules, initiating electron avalanches and leading to streamer formation under suitable conditions [8,9,10]. These early processes are relevant to the initiation phase of lightning, such as corona and streamer discharges [9,10,11].

The emission of light during these discharges is caused by electron collisions that excite atoms and molecules. When electrons return from excited states to their ground states, they emit photons at characteristic wavelengths, as represented in Figure 2a–c. Notably, the second positive system of nitrogen molecules (N₂ 2P), the first negative system of nitrogen ions (N₂⁺ 1N), atomic nitrogen (N), and atomic oxygen (O) contribute to visible and near-UV emissions [12,13,14,15,16]. Corona discharge in a laboratory setting exhibits higher peaks in the 300–400 nm range [15,17] with the emission of N₂ (2P) [17]. Lightning discharges exhibit distinct spectral signatures that reflect the underlying physical conditions of each phase. During the early stage of lightning, such corona discharges and streamers, the spectrum is dominated in the 300–450 nm range, particularly at 337 nm, corresponding to the N₂ (2P) [18]. In the latter process of lightning, stepped leader and return stroke, which is the strongest emission and energy, the spectrum broadens across 400–900 nm, initially showing strong ionized nitrogen (NII) and hydrogen Hα emissions, followed by enhanced emissions from neutral nitrogen (NI) and oxygen (OI), especially the OI triplet at 777.4 nm [19,20,21,22,23].

Conductivity also influences discharge morphology. In both air and water, higher conductivity tends to shorten and intensify discharge channels. Air has very low conductivity (~10⁻¹⁴ to 10⁻¹² S/m), but its conductivity increases with humidity [24,25]. In contrast, water’s conductivity varies widely depending on ion content: DI water is nearly insulating (~10⁻⁶ to 10⁻³ S/m), while tap water and salt solutions can reach ~0.05–5 S/m [26,27,28].

In this study, we investigate early lightning-like discharges using Tesla coil experiments as a simplified analog system. Our objective is to gain insight into the early stages of lightning development by analyzing two key aspects: (1) the optical emission spectra, which reveal information about the excitation processes and discharge composition, and (2) the spatial discharge morphology, which reflects how the discharges propagate under different environmental conditions such as humidity, surface type, and conductivity. By examining both spectral characteristics and spatial patterns, we aim to identify features analogous to natural lightning initiation, particularly corona and streamer behavior, and to understand how environmental conductivity affects discharge structure and emission properties. These insights contribute to a deeper understanding of spark discharge physics and their relevance to atmospheric electrical phenomena.

2. Materials and Methods

A mini-Tesla coil (Joytech, Fort Myers, FL, USA) was powered by a 48 V, 1.2 A DC supply (Figure 3a), producing sparks at approximately 100 kV and 50 kHz. A laundry steamer provided moist air (~70 °C) when required. Spark discharges were recorded with an iPhone camera (Apple Inc., Naples, FL, USA) at 30 fps (33 ms/frame) for qualitative visualization. Due to automatic exposure and color correction, these images were used solely to illustrate discharge morphology. Emission spectra were collected using a Spectryx USBVIS miniature CCD spectrometer (Spectryx Inc., Palo Alto, CA, USA; Figure 3b). Key specifications included a 1.3 nm FWHM resolution, 400–850 nm nominal range (typical: 350–950 nm), no order-sorting filter, 12-bit ADC, 40 ms–10 s integration time, and USB interface. A fiber optic cable (SMA-terminated) collected light from discharges and transmitted it to the spectrometer, which displayed and saved data via a Windows-based GUI.

To examine discharge spectra in air, the Tesla coil was mounted horizontally, with the electrode parallel to the tabletop (Figure 3c). The fiber tip was positioned 7 cm from the electrode, aligned horizontally and laterally offset by 1–2 cm to capture emissions from nearby spark branches without direct contact. Although spark directions varied randomly, some discharges propagated toward the fiber, resulting in high-quality spectra. In this configuration, discharges propagated only through air and did not interact with solid or liquid surfaces. To analyze discharges directed toward a water surface, the Tesla coil was mounted upside-down, with the electrode pointing vertically downward. The coil rested across two parallel, narrow sticks placed on adjustable-height stands (Figure 3d), allowing precise control of the electrode-to-water distance. The electrode tip extended slightly below the bottom of the sticks, maintaining stable vertical alignment. For spectral measurements, the electrode was set 4 cm above the water surface, a distance chosen to ensure most discharges propagated toward the water. The fiber tip was placed ~2 cm laterally from the discharge path. Spectra were collected continuously for ~20 s (approximately 400 spectra per trial). To reduce artifacts due to local air changes from repeated discharges, longer acquisition times were avoided. Spectra in which all wavelength intensities were below 45 were excluded; only those with at least one intensity value above 45 were retained for analysis.

Discharge lengths were measured using thermal paper under five atmospheric conditions: dry air (20 °C), moist air (~70 °C), heated air (~70 °C via heat lamp), headwind, and tailwind. The Tesla coil was placed upright on a table, with thermal paper positioned vertically behind it and a metal clip affixed midway on the right side (Figure 3e). As sparks were generated, the coil was moved from left to right. The longest distance at which a discharge visibly connected with the clip was defined as the maximum discharge path. The same upside-down coil setup (Figure 3d) used for vertical discharge was employed to determine the breakdown distance, namely the minimum height at which a continuous spark connected the electrode to the water, by gradually lowering the coil using the adjustable stands. Local electric fields during discharges were measured with an Erickhill EMF tester (Erickhill, Carlsbad, CA, USA) (range: 1–1999 V/m), which was placed at fixed distances from the electrode and water surface to evaluate field strength in different configurations. The conductivities of DI and tap water were measured using a PASCO PS-2116A conductivity probe with a PASPORT interface (PASCO scientific, Roseville, CA, USA). The software automatically displayed real-time conductivity.

3. Results

3.1. Spectral Analysis

Spark discharges generated by a Tesla coil were observed and recorded in the Physics Laboratory at Florida Gulf Coast University. Spectral data were collected for two different configurations: discharges in air, as detailed in Section 3.1.1, and discharges directed toward a water surface, as detailed in Section 3.1.2. The following subsections describe the results of spectral analysis for each case. Section 3.1.3 presents the comparison between the spectral data in Tesla coil discharges and different processes of natural lightning.

3.1.1. Spectrum of Discharge in Air

Figure 4a shows a typical image of spark discharges in air. Several sparks radiate outward from the electrode tip, forming multiple branches. The discharge exhibits a violet glow, especially bright near the tip. The arrow labeled “View (spectrum)” in Figure 4a indicates the direction in which the emission spectra were collected from the fiber tip. Figure 4b–e present four representative spectra recorded from individual discharge events. These were selected from a total of 61 valid spectral files based on the intensity and distinctness of their peaks. The spatial spread of the discharge allows emissions from various directions to reach the spectrometer, while temperature changes over time can alter the emitting species through chemical reactions. These factors likely contributed to the observed variations in both wavelength and intensity. To account for this variability, the study examines not only selected individual spectra but also the average and maximum spectral intensities derived from all spectral files. The peak wavelengths are labeled directly on the spectral plots in Figure 4b–e, where each value is positioned vertically next to the corresponding intensity maximum. Figure 4b displays a strong peak at 357.3 nm, representative of spectra dominated by emissions in the 350–380 nm range. Figure 4c shows a peak at 428 nm, typical of spectra with dominant emission between 400 and 450 nm. Figure 4d highlights a sharp peak at 892.2 nm, indicating near-infrared emission likely related to molecular oxygen or other minor species. Figure 4e presents a spectrum with multiple moderate peaks (at 366.4, 382.3, 392.2, 417.7, 491.7, 532, 702.6, and 795.2 nm), indicating the presence of several excited species simultaneously. The identification of emitting species corresponding to observed spectral peaks was based on wavelength matches with known atomic and molecular emission lines listed in the NIST Atomic Spectra Database [29]. Among the 61 recorded spectra, 5 exhibited peak intensities exceeding 400 within the 350–380 nm range, while another 5 displayed peaks greater than 500 in the 400–510 nm range. The peak wavelengths and their corresponding emitting species are summarized in

Table 1. Peak wavelengths and their corresponding emitting species. These peaks correspond to those shown in Figure 4b–e.

Peak Wavelength (nm)	357.3	366.4, 491.7	382.3, 417.7	392.2, 428	532	702.6, 795.2	892.2
Emitting species	N2 (2P)	NI	NI/N2	N2+ (1N)	NI/OI	ArI	OH

N₂ (2P): second positive system of neutral nitrogen molecules; N₂⁺ (1N): first negative system of nitrogen molecular ions; NI: atomic nitrogen (neutral); N₂: nitrogen molecule (neutral); ArI: neutral argon (neutral); OI: atomic oxygen (neutral); OH: hydroxyl radical.

. These spectra, shown in Figure 4b–e, include multiple strong emissions in the 350–450 nm region (peaks at 357.3 nm, 392.2 nm, and 428 nm), commonly associated with N₂ (2P) and N₂⁺ (1N) systems. Peaks at longer wavelengths, such as 702.6 nm and 795.2 nm, were also observed and attributed primarily to atomic Ar lines, likely introduced from ambient air or electrode interaction.

Figure 5a,b show the average and maximum spectral intensities at each wavelength, respectively, derived from all 61 valid spectral files. To aid interpretation, the prominent peak wavelengths are directly labeled on both plots. Notably, strong and consistent peaks appear between 350 and 450 nm, particularly at 357.3, 368.3, 403.8, and 428.4 nm, in both spectra. All labeled peak wavelengths are identical in the average and maximum spectra, highlighting the consistency of dominant spectral features despite pulse-to-pulse intensity variations. The wavelengths and corresponding emitting species for the spectra shown in Figure 5a,b are summarized in

Table 2. Peak wavelengths and their corresponding emitting species. These peaks correspond to those shown in Figure 5a,b.

Peak Wavelength (nm)	357.3	368.3	403.8	428.4	460.2, 489
Emitting species	N2+ (2P)	N2 (2P)/NI	NI/ArI	N2+ (1N)	ArI

. As in the individual spectra, prominent emissions were observed from the N₂ (2P) and N₂⁺ (1N) bands, along with several lines appearing beyond 500 nm.

3.1.2. Spectrum of Discharge Toward Water Surface

Figure 6a shows a spark discharge toward the water surface. While the sparks and branching structures exhibit a violet glow similar to air discharges, the primary discharge channel from the electrode tip to the water surface appears whitish violet. The discharge occurs intermittently, repeatedly turning on and off, with each pulse contacting a slightly different point on the water surface. Figure 6b–d illustrate representative single-peak spectra across these ranges, while Figure 6e shows an example featuring multiple distinct emission peaks distributed throughout the spectrum. A total of 64 spectral files were analyzed. Figure 6b shows a peak at 369.8 nm, typical of spectra with dominant emission between 350 and 380 nm. Figure 6c highlights peaks at 413.1 and 420.3 nm with a range of 400–420 nm. Figure 6d presents high peaks at 599.2 and 609.1 nm, and other peaks (at 381.5, 400.4, 434.5, 507.8, and 561.4 nm), and Figure 6e shows the peaks with a long range (at 350.1, 374.1, 557.1, 693.3, and 892.2 nm). Among these 64 spectra, 8 spectra showed peak intensities above 600 in the 360–450 nm range, another 8 had peaks above 500 in the 450–540 nm range, 4 spectra exhibited peaks above 350 in the 580–620 nm range, and 3 showed intensities above 250 in the 650–700 nm range.

Table 3. Peak wavelengths and their corresponding emitting species. These peaks correspond to those shown in Figure 6b–e.

Peak Wavelength (nm)	350.1, 374.1	369.8	381.5	413.1, 400.4	420.3	434.5	507.8, 561.4, 577.1, 599.2, 609.1, 693.3	892.2
Emitting species	N2 (2P)	NI/ArI	N2+ (1N)	NI/OI	NI	H$\mathsf{\gamma }$	ArI	OH/H2O+

H$\mathsf{\gamma }$: hydrogen Balmer gamma line.

represents the peak wavelengths and their corresponding emitting species.

Figure 7a displays the average intensity spectrum across all valid data. While the dominant peaks remain in the 380–480 nm range (with peaks at 381.1 nm, 389.6, 421.1 nm, and 479 nm), additional emissions are clearly observed at longer wavelengths, particularly around 599.2 nm and 609.6 nm. Figure 7b presents the maximum intensity spectrum across all 64 files. Strong emissions again appear in the 350–550 nm region (peaks at 378.2 nm, 420.3 nm, 478.6 nm, and 500.4 nm), as well as in the 600–620 nm range (at 609.1 nm).

Table 4. Peak wavelengths and their corresponding emitting species. These peaks correspond to those shown in Figure 7a,b.

Peak Wavelength (nm)	381.1	378.2, 478.6	389.6	420.3, 609.1, 609.6	421.1	479, 500.4, 516.6	507.3	599.2
Emitting species	NI/N2	N2+ (1N)	H$\epsilon$	NI	N2+	ArI	OI/ArI	NI/OI

H$\epsilon$: hydrogen Balmer epsilon line.

represents the peak wavelengths and their corresponding emitting species. Although many spectral features resemble those observed in air-only discharges (Section 3.2.1), a notable difference in the air-to-water-surface discharges is the appearance of additional peaks in the 500–650 nm range. This range includes emissions from species such as Ar, NI, and OI, as indicated in Table 3 and Table 4. These emissions are likely enhanced by various processes occurring at the plasma–liquid interface. For example, atomic recombination occurs when previously ionized atoms near the surface recombine, releasing photons as they return to a neutral state. Molecular excitation involves high-energy electrons exciting gas molecules such as N₂ and O₂, which subsequently emit light upon relaxation. Additionally, secondary reactions involving dissolved gases may take place, where radicals or electrons generated by the discharge interact with dissolved O₂, N₂, or Ar, forming excited species that contribute to the observed spectra [30,31]. These mechanisms likely account for spectral features that are absent in air-only discharges. These longer-wavelength emissions, which are less prominent in air-only discharges, may correspond to discharge processes that occur in later stages of lightning, such as the stepped leader approaching the ground or early pre-return-stroke activity at the surface.

3.1.3. Comparison Between the Spectrum of Tesla Coil Discharges and Lightning Emissions

The spectral characteristics of Tesla coil spark discharges were compared with those of natural lightning processes, focusing on emission sources and wavelength ranges.

Table 5. Characteristics of lightning discharges and Tesla coil discharges.

Processes	Main Emission Source
Return Strokes [19,20]	OI, NII, NI, H$\alpha$
Stepped Leader [21,22,23]	OI, NII, NI, H$\alpha$
Corona/Streamer [18]	N2 (2P)
Tesla Coil Discharge into Air (Our Study)	N2 (2P), N2+ (1N)
Tesla Coil Discharge toward Water Surface (Our Study)	N2 (2P), N2+ (1N)

summarizes the main emission sources associated with each process. Although weak lines of ArI, NI, and OI were also detected in the Tesla coil spectra, they were excluded from the table due to their relatively low intensities. Only the most prominent emissions were listed to maintain consistency across all discharge types. As described in Table 5, discharges in air showed strong emissions from N₂ (2P), closely resembling those observed in corona and streamer discharges in lightning. This suggests that Tesla coil sparks can serve as analogs for early-stage lightning phenomena. In addition, N₂⁺ (1N) bands were also observed in the Tesla coil spectra. These emissions are typically associated with more advanced ionization stages and may reflect localized regions where streamer-to-leader transition conditions are approached. In contrast, stepped leaders and return strokes exhibit different spectral signatures, with prominent lines from OI, NI, NII, and Hα, reflecting higher-temperature plasma and more complete ionization. Thus, while some overlap exists, Tesla coil discharges differ fundamentally in their dominant species and excitation mechanisms.

3.2. Spark Discharge in Various Air Conditions

In this section, we examine how atmospheric conditions, particularly moisture, affect Tesla coil spark discharges. Section 3.1 presents a comparison of discharges toward a water surface under dry and moist air environments to observe differences in discharge behavior. Section 3.2 introduces the use of thermal paper to measure discharge lengths under various atmospheric conditions. Section 3.3 reports electric field measurements at distances in both dry and moist air, providing additional physical context for the observed discharge phenomena.

3.2.1. Spark Discharge in Air Versus Moist Air

Spark discharges were generated between a vertically mounted electrode and the surface of tap water under two ambient conditions: room-temperature air (~20 °C) and moderately humid air (~70 °C) near a steam outlet. The temperature in the humid condition is approximate, as the warm, moist air mixed with the surrounding atmosphere. In both environments, whitish-violet sparks with violet branching were observed extending from the electrode tip toward the water. In room air, multiple sideways branches appeared (Figure 8a), while in humid air, fewer and more focused branches were observed (Figure 8b). Sparks propagated downward from the electrode in both cases. However, a continuous connection to the water surface only occurred when the tip was lowered below a certain height. We defined this critical height as the breakdown. Initial comparisons using tap and DI water showed no significant difference in distance, suggesting water conductivity had little influence under these conditions. Thus, tap water was used in all trials to focus on the role of air humidity. The average height at which the first spark reached the water surface was 8.9 cm in room air and 7.6 cm in moist air, based on ten measurements for each condition. These average values were taken as the breakdown distances and used to estimate the corresponding breakdown electric fields. The breakdown electric field E was estimated using the relation E = V/d, where V is the peak output voltage and d is the measured breakdown distance. Assuming a peak voltage of approximately 160–200 kV based on the Tesla coil settings (maximum voltage, minimum frequency), the corresponding electric fields were estimated to be approximately 1.8 × 10⁶–2.2 × 10⁶ V/m in dry air and 2.1 × 10⁶–2.6 × 10⁶ V/m in humid air (see

Table 6. Conductivity of dry and moist air, average breakdown height, and electric breakdown field.

Air	Conductivity (S/m)	Breakdown Height (cm)	Electric Breakdown Field (V/m)
Dry air	1.0 × 10−14 [24]	8.9	1.8 × 106–2.2 × 106 V/m
Moist air	1.0 × 10−12 [25]	7.6	2.1 × 106–2.6 × 106 V/m

). These results indicate that the presence of moisture lowers the breakdown threshold, allowing spark initiation at a shorter distance and under a weaker electric field.

3.2.2. Discharge Path Length in Various Air Conditions

To verify the findings from Section 3.2.1, discharge lengths were recorded using thermal paper under several atmospheric conditions: dry air, moist air (as in Section 3.2.1), heated air, and both headwind and tailwind airflow. The maximum discharge length was defined as the distance from the electrode tip to the metal clip when the discharge visibly reached the clip. As shown in

Table 7. Average length of the discharge path in various air conditions.

Air Conditions	Dry Air	Moist Air	Heated Air	Headwind	Tailwind
Average discharge path (cm)	6.26	5.29	6.62	6.30	6.22
Standard deviation (cm)	0.002	0.004	0.005	0.005	0.004

, the average maximum discharge path in air, moist air, heated air, headwind, and tailwind was 6.26 cm, 5.29 cm, 6.62 cm, 6.30 cm, and 6.22 cm, respectively. In a comparison of the lengths of the discharge paths in air, headwind, and tailwind, they show similar lengths. The average path in heated air is 0.36 cm longer than in air, while the average path in moist air is 0.97 cm shorter than in air. Interestingly, the discharge path in moist air is noticeably shorter compared to that in the other conditions. One possible explanation is that the smaller amount of induced charge at the metal clip due to humidity in the moist air reduces the discharge path length, as the reduced induced charge may prevent the discharge from propagating fully. Although the steam-generated moist air is warm and slightly turbulent, comparisons with heated air and airflow conditions indicate that humidity, not temperature or airflow, is the main factor reducing discharge length. As noted in Section 2, the temperature of the moist air near the outlet is only an estimate, and precise humidity control was not available. Future experiments will employ a controlled environmental chamber to address this limitation.

3.2.3. Electric Field Measurements of Discharge in Air Versus Moist Air

Electric field measurements were performed to compare the field strength generated by Tesla coil discharges in dry and moist air. Water was placed beneath the electrode tip to initiate the discharge, and the electric field was measured at horizontal distances ranging from 5.0 cm to 50 cm, aligned with the water surface. While moist air surrounded the discharge region, all sensors remained in unaffected dry air. Baseline electric fields without discharge were measured under both conditions, and these values were subtracted from the discharge measurements to isolate the contribution from the discharge itself. Due to the unstable nature of Tesla coil discharges, field readings fluctuated; therefore, average values were used for analysis. It is important to note that the EMF meter was used to compare relative field strengths under different ambient conditions, not to determine breakdown thresholds.

Table 8. Electric field at various distances from the discharge source in air and moist air, with average and standard deviation.

Distance (m)	Electric Field Due to Spark Discharge in Air (V/m)	Standard Deviation (V/m)	Electric Field Due to Spark Discharge in Moist Air (V/m)	Standard Deviation (V/m)
5	1018	102	903	67
10	719	101	598	24
15	499	30	255	28
20	241	0.6	98	7.2
25	144	1.0	81	14
30	115	3.2	88	8.9
35	109	4.4	95	5.3
40	105	2.0	89	2.6
45	98	2.1	121	27
50	93	4.0	94	2.1

presents the electric field values measured at distances ranging from 5.0 cm to 50 cm for spark discharges in both air and moist air. Each value represents the average of three measurements with corresponding standard deviations. Figure 9 shows scatter plots of the electric field versus distance, based on the data from Table 8. At 5.0 cm, the electric field is 1018 V/m in air and 903 V/m in moist air. Both conditions exhibit similar trends: the electric field is strongest near the source and decreases rapidly at an almost constant rate up to 25 cm. Beyond this point, the decrease continues more gradually. Across the entire range, the electric field in air remains consistently higher than in moist air. However, at 50 cm, the values converge, measuring 93 V/m in air and 94 V/m in moist air, becoming nearly identical.

3.3. Discharge over Various Water Surfaces

We investigate how surface conductivity and discharge height affect the spatial extent of star-shaped spark patterns. In Section 3.3.1, a comparison between star-shaped discharges over tap water and those over DI water is presented; in Section 3.3.2, the influence of NaCl concentration on star-shaped discharges over DI water is examined, and in Section 3.3.3, the relationship between spark discharge height and the resulting star-shaped path length is investigated.

3.3.1. Star-Shaped Discharge over Tap Water Versus DI Water

Spark discharges were generated from an electrode tip toward two water surfaces: tap water and DI water. Figure 10a,b show the overall discharge patterns, while Figure 10c,d provide zoomed-in views for tap water and DI water, respectively. Under identical laboratory conditions, DI water conductivity was 4.88 × 10⁻⁴ S/m, and tap water conductivity was 3.76 × 10⁻² S/m (21.6 °C). These values characterize the conductive properties affecting discharge behavior. On both surfaces, discharges spread radially into star-shaped patterns; however, the radius is smaller, and the channels are thicker over tap water compared to DI water. Moreover, when the electrode tip was positioned close to the tap water surface, spark initiation was delayed by several tens of seconds, indicating slower air ionization over tap water.

Table 9. Conductivity of DI water and tap water, and average maximum discharge height with standard deviation.

Surface Material	Conductivity (S/m)	Max Discharge Path Length (cm)	Standard Deviation (cm)
DI water	4.88 × 10−4	1.07	0.06
Tap water	3.75 × 10−2	0.429	0.03

summarizes the conductivities and the maximum radial path lengths measured with the electrode tip 1 cm above each surface. Discharge spread was defined as the radial distance from the electrode tip to the outer edge of the pattern, averaged over ten trials, with standard deviations reported.

3.3.2. Star-Shaped Discharge over DI Water with NaCl

We investigated star-shaped discharges on DI water with NaCl concentrations of 0.01%, 0.1%, 0.2%, 0.3%, 0.4%, 0.5%, 1%, 2%, 3.5%, and 5%.

Table 10. NaCl concentration in DI water, measured conductivity, and spark discharge appearance.

NaCl Concentration (%)	Conductivity (S/m)	Spark Appearance
0.01	2.41 × 10−2	Star-shaped
0.1	4.32 × 10−2	Dot-like
0.2	2.51 × 10−1	Dot-like
0.3	4.38 × 10−1	Dot-like
0.4	6.34 × 10−1	Dot-like
0.5	8.14 × 10−1	Dot-like
1.0, 2.0, 3.5, 5.0	-	Dot-like

Note: Conductivity values were measured for NaCl concentrations up to 0.5%. Higher concentrations were not measured due to limitations in instrument sensitivity or experimental conditions.

summarizes the NaCl concentration, measured conductivity (up to 0.5%), and observed discharge patterns. Note that the conductivity sensor’s upper limit is 1.0 S/m, so higher concentrations could not be measured. Except at 0.01%, all NaCl concentrations produced dot-like discharges with no radial spread, consistent from 0.1% upward. At 0.01%, discharges retained a star-shaped pattern and spread area similar to pure DI water. The dot-like behavior at 3.5% (approximating seawater) further confirms that high conductivity suppresses radial propagation. Although tap water also exhibited a star-shaped pattern, its spread area matched that of DI water with a low NaCl concentration.

3.3.3. Spark Discharge Height Versus Star-Shaped Discharge Path Length

Star-shaped discharges over tap water and DI water were recorded at electrode-tip heights from 7.0 cm down to 1.0 cm in 1.0 cm increments (Figure 11, top row). A 5 min pause between measurements ensured consistent ambient conditions. As height decreased, the number of violet-branched filaments diminished, while the primary whitish discharge to the DI water surface thickened. The middle row of Figure 11 shows zoomed-in images of the DI water discharge path. Here, lower electrode heights produced a wider star-shaped spread and larger regions of intense luminosity. The bottom row of Figure 11 maps luminosity intensity (red = highest intensity), confirming that the high-intensity area expands along the discharge path as the tip approaches the surface. Additionally, the discharge color shifts toward whitish hues at lower heights, suggesting enhanced air ionization and stronger discharge.

Table 11. Discharge height and average discharge path lengths over DI water and tap water, with standard deviations.

Height (cm)	Average Radial Path over DI Water (cm)	Standard Deviation (cm)	Average Radial Path over Tap Water (cm)	Standard Deviation (cm)
7	0.696	0.07	0.379	0.002
6	0.724	0.05	0.384	0.02
5	0.796	0.05	0.400	0.03
4	0.896	0.07	0.413	0.008
3	0.917	0.07	0.468	0.009
2	0.942	0.07	0.421	0.01
1	1.067	0.06	0.429	0.03
0	1.047	0.02	0.443	0.006

lists the average radial path lengths (±standard deviation) over DI water and tap water for each initial height for ten measurements. Figure 12 plots these averages against electrode-tip height. For DI water, the spread length increases with decreasing height, showing a strong linear correlation (R² = 0.96). For tap water, the relationship is weaker but still positive (R² = 0.59).

4. Discussion

Spectral analysis revealed that Tesla coil discharges are dominated by N₂ (2P) emissions, which are characteristic of non-thermal corona and streamer discharges. These emissions are primarily confined to the 350–450 nm range. This spectral confinement closely aligns with that of early-stage lightning phenomena, such as corona and streamer discharges, which typically exhibit N₂ (2P) features between 390 and 440 nm [18]. In contrast, later-stage lightning processes, such as stepped leaders and return strokes, produce broader emission spectra spanning 300–1000 nm and include high-temperature lines such as Hα and OI [32], which are absent in the Tesla coil discharges. Despite being generated by high-frequency AC, the Tesla coil sparks thus replicate key features of early lightning activity. Furthermore, the detection of N₂⁺ (1N) bands suggests that the electric fields in these discharges may be strong enough to induce ionization beyond neutral excitation, potentially resembling the streamer-to-leader transition in natural lightning. This observation supports the view that ionization is essential for dielectric breakdown in the atmosphere, as exemplified by lightning [33], and that spectral characteristics can be influenced by variations in atmospheric molecular composition and concentration [2].

Environmental factors played a notable role in discharge behavior. Low humidity increased branching and propagation, while high humidity favored shorter but more easily initiated sparks. Minor effects from airflow and temperature were ruled out, confirming that humidity itself plays a key role, likely through ion recombination and the presence of water droplets. Surface conductivity had a pronounced effect: discharges over DI water extended farther than those over tap water or salt water. In low-conductivity media, electric fields persist due to reduced current flow, interpreted through Ohm’s law, J = σE, where J is the current density, σ is the electrical conductivity, and E is the electric field, allowing discharge growth; in contrast, conductive surfaces dissipate fields quickly, suppressing propagation. Interestingly, the morphology of Tesla coil discharges also reflected distance-dependent effects. Shorter electrode–surface gaps produced intense, radially spreading discharges, likely due to stronger local electric fields and increased vaporization at the interface. These findings may offer insight into lightning behavior over ground versus ocean surfaces, where conductivity and topography influence channel development.

Though Tesla coil discharges differ from natural lightning in waveform and scale, they can replicate localized electric field conditions relevant to streamer and corona formation. This opens the possibility of using such systems to model upward lightning initiation, under the assumption made in some studies that corona discharges occur in its early stages [34,35,36] but also to simulate streamer formation at the point where an upward connecting leader from the ground or a ground-based object attaches to a downward leader [37]. Our results also suggest that weak, upward-directed streamer or corona discharges could originate from the ground in response to approaching downward leaders, particularly over poorly conductive terrain. This may influence the attachment process in ways not fully captured by current models.

Although our spectra were recorded as wavelength–intensity plots rather than direct color images, they represent the visible light emitted by spark discharges, similar in nature to lightning emissions. To symbolically illustrate this connection, we include in Appendix A a photograph taken by the first author, G. Steinberg, that captures both a CG lightning discharge and a rainbow. While these phenomena originate from different physical processes, electrical breakdown and optical dispersion, they are both atmospheric light events governed by interactions of radiation with air and water. Their visual coexistence highlights the spectral richness of lightning and the relevance of optical diagnostics in atmospheric electricity research.

5. Conclusions

This study investigated spark discharges generated by a Tesla coil under varying humidity and surface conductivity conditions. Optical spectral analysis revealed that the discharges are dominated by N₂ (2P) emissions, with occasional detection of N₂⁺ (1N) bands. These emissions are primarily confined to the 350–450 nm range, closely aligning with those observed in early-stage lightning phenomena such as corona and streamer discharges. The morphology and spread of the discharges were strongly influenced by the conductivity of both the surrounding air and the surface. Low-conductivity environments, such as dry air and DI water, facilitated longer, more complex, and radially expanding discharges. In contrast, high-conductivity surfaces, including tap water and saltwater, suppressed lateral propagation and resulted in more compact spark structures. Humidity also played a significant role: dry conditions supported higher breakdown thresholds and longer spark propagation. Additionally, the gap between the electrode tip and the surface affected discharge geometry, likely through localized field enhancement and surface vaporization effects. Although the experimental conditions differ from natural lightning in terms of polarity, scale, and waveform, the results offer useful analogies for early lightning processes. Future studies employing larger-scale, pulsed-DC, or more lightning-representative setups will be essential to further investigate lightning initiation and propagation under diverse atmospheric and surface conditions.