1 Introduction

Aerosols present in the atmosphere vary in size, shape, and source of origin. Notably, biological particles (bioaerosols), including pollen, microorganisms, and artificially manufactured toxic substances, exist to the extent of 5 to 30% in micron sizes. Particles of biological origin stem from all activities of life, environment, industry, and the interplay between humans and nature. Among which, bioweapons and biological terrorism substances are primarily engineered to be 2 to 10 µm in size for easy penetration through the human respiratory system (Jeong et al., 2014; Layshock et al., 2012; Primmerman, 2000).

Biological terrorism substances can be lethal even in very low concentrations, capable of proliferating upon infecting a host, hence possessing a high fatality rate. Particularly, their release in confined areas can, through dispersion, affect a large population over a wide area in a short span, making them a significant threat (Graham et al., 2008). Furthermore, they can cause long-term adverse effects on numerous individuals, not only impacting health but also the living environment (Kim et al., 2009; Lighthart & Stetzenbach, 1994), and the invisibility of these weapons exacerbates fear, leading to social unrest. The advancement in biotechnology, enhancing the toxicity and transmissibility of biological weapons and facilitating their mass production, renders the possibility of bioterrorism an issue that cannot be ignored.

Bioweapons were used during World War I for dispersing anthrax and in large-scale field experiments on war prisoners during World War II, intensifying anxieties. However, these stabilized after the Biological and Toxin Weapon Convention (BTWC) in 1972. Nonetheless, the importance surged again following the anthrax-based postal bioterrorism on October 4, 2001, which resulted in 22 casualties, and the possibility of biological terror attacks through aerosol routes is considered (Kamboj et al., 2006).

To rapidly detect and prevent the spread of biological terrorism, monitoring technologies capable of swiftly detecting the presence of biological warfare aerosols from a distance, and more precisely, analyzing changes in density to provide information on the spread rate for tactical decisions, are necessary.

While in situ measurement techniques for monitoring biological particles are already developed and in use (Farka et al., 2015; Shen et al., 2011), these techniques cannot offer information on spatial distribution and spread due to their limited measuring range. They necessitate numerous measurement points to monitor extensive areas, which is not ideally suited for military purposes.

The realistic and available remote sensing technology capable of providing information on spatial distribution is Light Detection and Range (LiDAR) technology. LiDAR utilizes laser as a light source to offer physical and chemical information alongside distance data for the target element. Fluorescence LiDAR technology, exploiting the fluorescent characteristics of biological particles, has been developed (Joshi et al., 2013; Mejean et al., 2004). However, due to the weak intensity of fluorescence, daytime observations are impossible, presenting a limitation for continuous monitoring. A solution is the use of elastic scattering LiDAR, which can continuously operate 24/7, allowing the identification of biological particles by confirming their size and shape, thus detecting their presence.

This study performed a simulation using elastic scattering LiDAR before actual observation using a chamber capable of releasing biological particles. It calculated the wavelength-specific extinction coefficients at applicable wavelengths (266, 1064, 1571, and 2000 nm) and derived the Ångström exponent (AE) from these coefficients to provide size information of the particles, thereby verifying the distinguishability between general aerosols and biological particles.

The selection of simulation wavelengths was driven by the need to detect bioaerosols as coarse particles in the atmosphere while excluding the 532 nm wavelength due to eye safety issues. Wavelengths within the visible spectrum pose eye safety risks and would not be feasible for use in populated areas should bioaerosol dispersal occur. Instead, infrared (IR) wavelengths of 1064, 1571, and 2000 nm were chosen as test wavelengths. These IR wavelengths are generally less efficient for fine particles due to lower scattering efficiencies but can potentially detect larger particles. Until now, aerosol observation using lasers at wavelengths of 532 nm or 355 nm has been common. However, recent efforts have been made to explore the suitability of 1064 nm wavelength lasers for aerosol observation (Chen et al., 2023; Li et al., 2023) and to use 1550 nm wavelength lasers for observing aerosols and clouds (Yue et al., 2022). These efforts aim to address the longstanding eye safety issues associated with LiDAR systems. Additionally, the 266 nm wavelength was selected for the detection of fluorescence in biological particles. Considering equipment design that can accommodate up to four wavelengths, the commonly used UV wavelength of 355 nm for aerosol detection was excluded from this study.

2 Remote monitoring simulation method for bioaerosols

The method for remotely monitoring biological particles involves indirectly measuring limited information and probabilistically determining how closely airborne particles match this information based on their proximity. The fundamental information distinguishing biological particles, with a size distribution of 2–10 µm in diameter, from general aerosols, is data on particle size. This study aimed to verify the extinction coefficient of biological particles at wavelengths applicable to elastic scattering LiDAR, using size distribution and refractive index information confirmed from preliminary research.

2.1 Calculation of extinction coefficient by Mie theory

The extinction coefficient serves as an important measure of air quality, reflecting the degree of light scattering and absorption by aerosols. The calculation of the aerosol extinction coefficient is influenced by factors such as the extinction efficiency (\(Q\)) and the aerosol surface size distribution (Balzarini et al., 2015; Bohren and Huffman, 2008; Myhre et al., 2013; Wang et al., 2014). From Eq. 1, we can calculate the theoretical extinction coefficient on specific wavelengths and aerosols ranging in certain sizes.

$$\alpha \left(\lambda ,m\right)={\int }_{0}^{\infty }Q\left(\lambda ,r,m\right)\pi {r}^{2}N\left(r\right)dr={\int }_{0}^{\infty }Q\left(\lambda ,r,m\right)\frac{3}{4}\frac{1}{r}V\left(r\right)dr$$
(1)

Here, \(\alpha\), \(\lambda\), and \(r\) are the extinction coefficient, wavelength, and the radius of aerosols, respectively. \(m\) means the refactive index and was assumed to be 1.45 + 0.002i for all cases (Veselovskii et al., 2004). \(N\left(r\right)\) is the number size distribution, and \(V\left(r\right)\) is the volume size distribution.

The volume size distribution can be explained by log normal distribution (Eq. 2). There are six parameters: volume, median radius, and the width of fine- and coarse-mode aerosols (\({V}_{f}\), \({V}_{c}\), \({r}_{f}\), \({r}_{c}\), \(\text{log}({\sigma }_{f})\), \(\text{log}({\sigma }_{c})\), respectively). In this study, we simulated volume size distribution depending on volume and median radius of aerosols.

$$\frac{dV(r)}{dlog(r)}={\sum }_{k=f,c}\frac{{V}_{k}}{\sqrt{2\pi }(\text{log}\left({\sigma }_{k}\right))}{e}^{-\frac{{(\text{log}\left({r}_{k}\right)-\text{log}\left(r\right))}^{2}}{2{(log{\sigma }_{k})}^{2}}}$$
(2)

We considered systems of various wavelengths for the discrimination of biological particles. We are conceptualizing a LiDAR system that includes elastic signals in the 266 nm wavelength region capable of fluorescence detection, along with elastic signals in infrared wavelength bands (1064 nm, 1571 nm, and 2000 nm). It is anticipated that at least two lasers will be equipped, and changes in the extinction coefficient and AE are examined to rapidly detect biological particles.

The AE is a parameter related to particle size; it tends to have values close to 0 when larger particles are present in the atmosphere and high values between 2 and 4 when smaller particles are present (Angström, 1929). The calculation of the AE can be represented by the following equation.

$$AE=-\frac{\text{log}(\alpha \left({\lambda }_{1}\right)/\alpha \left({\lambda }_{2}\right))}{\text{log}({\lambda }_{1}/{\lambda }_{2})}$$
(3)

\(AE\): Ångström exponent

\(\alpha\): extinction coefficient of aerosol

\(\lambda\): wavelength

2.2 Bioaerosol simulation

To efficiently design a LiDAR system capable of swiftly detecting the release of biological particles from a distant point, we conducted simulations under assumed conditions of biological particle release. The observed wavelengths were 266, 1064, 1571, and 2000 nm, with a comparison wavelength set at 532 nm. While 532 nm LiDAR is widely used for aerosol observation due to its excellent scattering efficiency, its military use is limited due to eye safety issues and the visibility of green light. Therefore, the purpose of this simulation was to identify wavelength regions capable of biological particle discrimination at a level that could substitute the 532 nm wavelength.

The simulation of biological particle release considered variations in the quantity and size of particles under atmospheric conditions (clean, normal, bad) (Table 1). Atmospheric conditions assumed quantities of particles (\({V}_{f}\) and \({V}_{c}\)) that meet the environmental standards set by the Ministry of Environment. Under PM2.5 standards, clean, normal, and bad conditions were defined as less than 15 µg m-3, 15–35 µg m-3, and over 35 µg m-3, respectively. Under PM10 standards, they were defined as less than µg m-3, 45–75 µg m-3, and over 75 µg m-3, respectively. The volumes of fine and coarse particles (\({V}_{f}\) and \({V}_{c}\)) that met these criteria were 8 and 15 µm3 cm-3 for clean, 20 and 30 µm3 cm-3 for normal, and 40 and 60 µm3 cm-3 for bad conditions. The median radius of fine and coarse particles (\({r}_{f}\) and \({r}_{c}\)) were assumed to be 0.15 µm and 2.8 µm, respectively, with distribution widths (\(\text{log}({\sigma }_{f})\) and \(\text{log}({\sigma }_{c})\)) assumed to be 0.5 and 0.7, respectively, which are mid-values for general atmospheric particles (Veselovskii et al., 2004). Using these parameters, the mass concentrations from volume size distribution were calculated to check atmospheric conditions, assuming an aerosol density of 1.5 \(\text{g} {\text{cm}}^{-3}\), which is the average density for the accumulation mode (Kannosto et al., 2008). The reason for simulating three different atmospheric conditions is that this study aims to immediately detect coarse-size bioaerosols under any background condition, potentially released in large amounts during a bioattack. Designing this experiment raised concerns that once bioaerosols are dispersed in the atmosphere, it would be challenging to distinguish them from general atmospheric particles based on size alone. It could be difficult to discern whether increases in particle concentration were due to bioaerosols or other sources such as dust storms, pollen, or pollution emissions. Therefore, we sought to verify the effectiveness of detecting specific sizes of bioaerosols even in conditions with high levels of atmospheric pollutants.

Table 1 Simulation conditions for atmospheric states and bioaerosol particles
Full size table

We examined changes in the extinction coefficient and AE under three atmospheric conditions based on particulate matter concentration when biological particles were released. The bioaerosol release scenarios considered include a change in the quantity of biological particles (volume of bioaerosol: 0–100 µm3 cm-3) and a case where a large amount of biological particles were released at a “bad” level (volume = 60 µm3 cm-3) with a change in particle size (medium radius of bioaerosol: 1–5 µm). The particle size was set considering that particles used for biological terrorism are typically engineered to have diameters of 2 to 10 µm for easier absorption through the human respiratory system. The bioaerosol simulation aimed to determine if specific changes in the quantity of particles could be detected over a wide area when a gradual release of actual biological particles occurs, and because biological particles are experimentally generated, whether the detection of specific size particles is as expected.

Clean-1 and Clean-2 cases involve changes in the quantity and median radius of particles when atmospheric pollution is at a good level. Normal-1 and Normal-2 correspond to moderate levels of atmospheric pollution, and Bad-1 and Bad-2 represent bad levels, with respective changes in volume and median radius. The changes in volume size distribution for each scenario are illustrated in Fig. 1.

Fig. 1
figure 1

Volume size distribution simulation results; a Clean-1, b Clean-2, c Normal-1, d Normal-2, e Bad-1, and f Bad-2 in Table 1

Full size image

3 Result and discussion

3.1 Simulation results in different wavelength

This study aimed to verify the detection efficiency of specific wavelengths beyond the visible spectrum through simulations assuming the release of biological particles in the atmosphere. The simulations classified the scenarios into three cases based on the mass concentration of background aerosols. Firstly, in the clean case, the PM10 and PM2.5 mass concentrations were 23.8 µg m-3 and 11.8 µg m-3 respectively, and the extinction coefficient at a wavelength of 532 nm was 0.54 km−1. In the normal case, the PM10 and PM2.5 mass concentrations and the extinction coefficient (at 532 nm) were observed to be 52.5 µg m-3, 28.3 µg m-3, and 0.129 km−1, respectively. In the scenario simulating a bad case, these values were calculated to be 105.0 µg m-3, 56.6 µg m-3, and 0.258 km−1, respectively. These conditions for the ambient environment were fixed into these three categories as they represent typical situations.

We investigated the release of bioaerosols under various ambient conditions, focusing on the volume of released particles and their size variations. It was posited that the quantity of particles during size variation scenarios would be equivalent to that observed in the bad case condition. Figure 2 illustrates the modeled variations in the extinction coefficients across different wavelengths as the volume and radius of biological particles change under each atmospheric scenario. This model aimed to ascertain the rate at which the extinction coefficient shifts when atmospheric particles are uniformly distributed.

Fig. 2
figure 2

Extinction coefficient (km−1) simulation results: a and b represent the clean case, c and d depict the normal case, while e and f illustrate the bad case

Full size image

Considering the widely used wavelength of 532 nm for aerosol detection, we aimed to assess the detection at other wavelengths, taking into account eye safety for system construction. Variations in the quantity of bioaerosols showed consistent changes in the extinction coefficient across clean, normal, and bad atmospheric conditions, with no significant differences observed across wavelengths. This indicates that changes in the extinction coefficient due to a rapid increase in bioaerosol quantity can be detected at various wavelengths. However, when particle size varied, with the quantity remaining constant (volume = 60 µm3 cm-3), an increase in particle size led to a decrease in extinction efficiency according to Mie theory, consequently reducing the extinction coefficient. This suggests that larger bioaerosol particle sizes result in lesser detectability of changes in the extinction coefficient. Minimal changes in the extinction coefficient were observed for particle sizes greater than 2, implying a central size of 2 µm, similar to the central size of coarse particles in the general atmosphere (2.2 µm). Hence, it can be deduced that distinguishing bioaerosols becomes difficult if they are dispersed at sizes akin to those of coarse particles.

Upon examining the variations across different wavelengths, changes in the extinction coefficient at 266 nm appeared to be significant. However, the short wavelength poses challenges for long-distance measurements, though it holds promise for use in night-time and fluorescence signal discrimination. For aerosol detection, the wavelength that showed efficiency most similar to the advantageous 532 nm was identified as 1064 nm in the infrared spectrum.

To detect changes when biological particles are dispersed as large particles, it is necessary to observe variations in mass concentration concurrently. The results of scattering coefficient simulations using available LiDAR wavelengths showed that it was challenging to detect rapid changes in the extinction coefficients due to variations in particle size or quantity. However, when simulating the change in mass concentration due to the release of bioaerosols as coarse particles, an evident increasing trend in PM10 is observed as the particle quantity (i.e., volume) increases (Fig. 3). This trend is primarily because biological particles predominantly exist in sizes larger than the PM2.5 diameter range.

Fig. 3
figure 3

PM mass concentration (μg m−3) simulation results: a and b represent the clean case, c and d depict the normal case, while e and f illustrate the bad case

Full size image

The variability in particle size presented a more complex pattern of change. As particle size increases, PM2.5 decreases and then stabilizes, which occurs as the median radius of the bioaerosol peak gradually exceeds the PM2.5 cut-off size, moving beyond the 1.25 μm range. Since most coarse particles are encompassed within PM10 range, their level remains constant until the median radius exceeds 4, at which point the peak shifts outside of the PM10 size range, leading to a decrease in mass.

To enhance detection efficiency in real measurements, it is necessary to comprehensively assess the changes in scattering coefficients and mass concentration due to variations in the quantity and size of biological particles upon their release.

3.2 Simulation results for the AE

The simulation results into the variation of extinction coefficients by IR wavelength for LiDAR design suggest that lasers operating at a wavelength of 1064 nm may offer the most favorable conditions. However, distinguishing biological particles based on extinction coefficients alone proves challenging, necessitating at least two different wavelengths of laser to enable the calculation of the AE, a critical measure for determining particle size. Extinction coefficients at wavelengths of 355 nm and 532 nm or 532 nm and 1064 nm are commonly used in numerous studies for deriving the AE. The wavelength of 266 nm is excluded from calculations related to the AE due to its significant deviation in the wavelength range from the infrared, reflecting findings from research indicating that larger wavelength disparities lead to increased errors in AE calculation (Guo et al., 2021). In an effort to establish the wavelength range for IR, taking into account the need to exclude 532 nm due to eye safety concerns, the study proposes using the AE values for 532 nm and 1064 nm as benchmarks. This approach seeks to review the AE across alternative wavelengths to identify the wavelength combination that yields the highest detection efficiency.

Figure 4 displays the simulation results of the AE in three atmospheric conditions for increasing quantities of biological particles (Fig. 4a, c, and e) and for changes in particle size (Fig. 4b, d, and f), similar to Figs. 2 and 3. AE was estimated across wavelength combinations of 532 nm and 1064 nm, 1064 nm and 1571 nm, 1064 nm and 2000 nm, and 1571 nm and 2000 nm. When the quantity of biological particles increases, the AE tends to decrease. Among the wavelength combinations, the pair of 532 and 1064 nm showed the most rapid decrease in the AE with an increase in biological particle concentration, with the other combinations yielding similar outcomes. The change in the AE was most significant in clean atmospheric conditions and decreased as the atmosphere changed from normal to bad. This result, while expected, implies that the influence of biological particles diminishes as the concentration of background particles in the atmosphere increases.

Fig. 4
figure 4

AE simulation results: a and b represent the clean case, c and d depict the normal case, while e and f illustrate the bad case

Full size image

The analysis of the AE with changes in the size of biological particles while keeping the particle quantity constant reveals distinct patterns across different wavelength combinations. Specifically, the combination of IR wavelengths behaves in unexpected patterns; it exhibits a sharp increase in the AE as the radius of biological particles increases from 2 to 3 μm, after which the increase becomes gradual. This is a discrepancy in the theory that AE decreases when aerosol size is large. This pattern suggests a sensitivity to size changes within a specific range, likely due to the optical properties and scattering behavior of particles within these wavelengths.

Contrastingly, with other wavelength combinations, the AE initially decreases sharply as particle size increases, followed by an increase. This indicates a non-linear relationship between particle size and the AE, possibly reflecting complex scattering dynamics at these wavelengths. The lowest AE observed at combinations of 1064 nm and 1571 nm, 1064 nm and 2000 nm, and 1571 nm and 2000 nm were at particle sizes of 1.26–1.32, 1.41–1.46, and 1.77–1.83 μm, respectively. The 1571 nm and 2000 nm combination showed the greatest variation, implying enhanced sensitivity to size changes, which could be advantageous for detecting biological particles of those specific sizes in the atmosphere.

Considering the diversity in biological particle sizes and the difficulty in specifying them, a wavelength combination that exhibits a less pronounced change in the AE with size variation could be more practical for confirming the presence of biological particles. The 532 nm and 1064 nm combination quickly reaches a stable AE value with increasing particle size, suggesting utilization of this wavelength pair in diverse atmospheric conditions without requiring precise particle size information.

Upon reevaluation, the 1064 nm and 1571 nm combination, which shows minimal variation for particles larger than 2 μm, could offer a balance between sensitivity and practicality. It can potentially identify biological particles without necessitating exact size determination, which is beneficial given the variability in bioaerosol sizes. Thus, excluding 532 nm due to eye safety concerns, the 1064 and 1571 nm combination emerges as a potentially optimal choice for detecting the presence or absence of biological particles, balancing between sensitivity to size changes and operational practicability. This conclusion is based on the differential sensitivity of these IR wavelengths to changes in both the quantity and size of biological particles, as reflected in their impact on the extinction coefficient and the AE, respectively.

4 Summary and conclusion

This research has simulated a method for the rapid detection of atmospheric biological terrorism agents through the utilization of advanced infrared (IR) LiDAR technology. The critical need for swift and accurate identification of harmful bioaerosols underscores the importance of our findings for public health and safety. By simulating the extinction coefficient and AE across various wavelengths (266, 1064, 1571, and 2000 nm), we have identified that the 1064 nm and 1571 nm wavelengths are notably effective in distinguishing biological agents from ordinary atmospheric particles. This discovery paves the way for developing specialized LiDAR-based systems tailored for the detection of bioaerosols, helping our response to biological threats with enhanced accuracy.

However, while our results are promising, they stem from simulations and need empirical validation under actual atmospheric conditions. This validation is crucial for confirming the practical applicability of our findings. In addition, the simulation used only the size information of bioaerosols, which are close to coarse particles. However, since bioaerosols used in bioattacks might be dispersed in large quantities, a significant increase in coarse particles is expected, making this classification method reasonable for identifying bioaerosols. To more accurately verify particle type, nighttime observations using the 266 nm wavelength are necessary. While 266 nm is not suitable for daytime observations, it can detect fluorescence signals from bioaerosols at night, providing valuable validation data. Future investigations should not only focus on validating these results but also on exploring the integration of additional wavelengths and advancements in LiDAR technology to broaden the scope of detection capabilities.

In conclusion, our study contributes significantly to the advancement of atmospheric monitoring techniques, introducing an effective strategy for the early detection and management of biological terrorism threats. The continued refinement, development, and empirical validation of this technology hold the potential to be instrumental in the global endeavors to minimize the hazards posed by the dispersal of bioaerosols, applicable in both civilian and military settings.