Short-Range High Spectral Resolution Lidar for Aerosol Sensing Using a Compact High-Repetition-Rate Fiber Laser

Abstract

This work presents a proof of concept for a short-range high spectral resolution lidar (SR-HSRL) optimized for aerosol characterization in the first kilometer of the atmosphere. The system is based on a compact, high-repetition-rate diode-based fiber laser with a 300 MHz linewidth and 5 ns pulse duration, coupled with an iodine absorption cell. A central challenge in the instrument’s development was identifying a laser source that offered both sufficient spectral resolution for HSRL retrievals and nanosecond pulse durations for high spatiotemporal resolution, while also being compact, tunable, and cost-effective. To address this, we developed a methodology for complete spectral and temporal laser characterization. A two-day field campaign conducted in July 2024 in Tsukuba, Japan, validated the system’s performance. Despite the relatively broad laser linewidth, we successfully retrieved aerosol backscatter coefficient profiles from 50 to 1000 m, with a spatial resolution of 7.5 m and a temporal resolution of 6 s. The results demonstrate the feasibility of using SR-HSRL for detailed studies of aerosol layers, cloud interfaces, and aerosol–cloud interactions. Future developments will focus on extending the technique to ultra-short-range applications (<100 m) from ground-based and mobile platforms, to retrieve aerosol extinction coefficients and lidar ratios to improve the characterization of near-source aerosol properties and their radiative impacts.

1. Introduction

Active remote-sensing instruments are a key component for atmospheric research, of which lidar (acronym of LIght Detection And Ranging) is one of the leading techniques, as it allows obtaining information about the presence [1], nature [2], and spatiotemporal distribution [3] of different atmospheric constituents covering ranges from the ground to >100 km of altitude. However, in recent years, there has been a growing need to understand and measure the impacts of aerosols near the source for various climate, air quality, and aeronautical applications. Thus, different types of short-range lidars have been developed to determine the radiative and microphysical properties of particles over short ranges.

The most common type of lidar is based on the elastic backscatter technique, where the wavelength is unchanged in the scattering process [4]. Based on the time-of-flight principle, laser pulses propagate through the atmosphere, and a photoelectric sensor detects the small percentage of light backscattered by the particles or gas molecules, providing range-resolved measurements of the aerosol distribution [4]. Short-range elastic backscatter lidars (SR-EBLs) have received growing attention from the scientific community. The micropulse lidar (MPL), a small-form factor low-power elastic backscatter lidar, has been widely used for cloud and aerosol detection. In 2021, Chen et al. [5] introduced the polarized MPL (P-MPL), enabling continuous observations of aerosol/cloud spatiotemporal variations, extinction coefficients, depolarization ratios, and planetary boundary layer (PBL) structure. The detection range of the P-MPL is 0.1 to 20 km, with a full overlap at ∼400 m. The spatial and temporal resolution is 37.5 m and 1 min. In 2020, Ceolato et al. [6] introduced a novel bi-static SR-EBL system characterized by its lightweight design, compactness, and compatibility with mobile platforms. The Colibri Aerosol Lidar (CAL) systems, designed for both vertical and horizontal operations, deliver quantitative aerosol profiling with <10 cm range resolution within the first hundredths of meters (from 10 m). To solve the EBL equation in the short range, where no reference molecular zone is available, a specific forward Fernald inversion method without boundary conditions was developed. This method involved performing geometric and radiometric calibrations, as well as characterizations of dark current and background noise. Later in 2022, Ceolato et al. [7] estimated the aerosol number and mass concentration of soot fractal aggregates at 8 to 10 m from the source using the CAL sensor. Also in 2020, Ong et al. [8] demonstrated the capability of a near-horizontal lidar to detect diurnal changes in aerosol extinction (at 349 nm) near the ground and to characterize the diurnal response of the aerosol extinction coefficient with respect to relative humidity. This near-horizontal lidar has an elevation angle of 4°, i.e., an observable range of ∼200 to 5000 m, equivalent to an altitude of approximately 350 m. It records the backscattered signal with a 5 min accumulation time at a 7.5 m range resolution. Recently, short-range coherent Doppler wind lidar (CDWL) measurements have been proposed for aerosol measurements from tens to hundreds of meters [9,10]. These studies reported that retrieving aerosol properties from coherent DWLs requires supplementary data and calibration with in situ measurements.

Despite their ability to characterize near-source aerosols with a high or ultra-high spatiotemporal resolution, the inherently ill-posed EBL problem limits the accuracy of all short-range elastic backscatter lidars, including MPLs, CALs, and CDWLs. This limitation comes as the signal $P\left(r,\lambda \right)$ measured at range r and wavelength $\lambda$ by an EBL is proportional to two physical quantities, the backscatter $\beta \left(r,\lambda \right)$ and extinction $\alpha \left(r,\lambda \right)$ coefficients. Thus, to solve the EBL equation, one must assume a linear relationship between extinction and backscatter, called lidar ratio (LR) and defined as $LR=\frac{\alpha \left(r,\lambda \right)}{\beta \left(r,\lambda \right)}$. Additionally, it is necessary to have extra atmospheric temperature and pressure profiles to calculate molecular extinction and backscatter. Due to these assumptions, assessing ultrafine particles with a backscatter coefficient close to the molecular value, such as freshly emitted soot, poses a significant challenge for elastic lidars. Inelastic backscatter lidars using Raman scattering bypass the EBL problem by directly measuring aerosol extinction and backscatter coefficients without relying on assumptions. However, these techniques generally require a long time resolution (typically hours) and usually operate under night conditions. The HSRL technique is another technique to solve the ill-posed EBL problem. It measures the Doppler shifts induced by molecules and aerosols. Depending on the atmospheric temperature and pressure, molecules move at ∼300 m/s by random thermal motion, inducing a frequency shift in the order of GHz (∼1 to 4). On the other hand, much more massive and slowly moving aerosols induce three orders of magnitude lower Doppler shifts (∼3 to 100 MHz) [11,12,13] depending on the laser wavelength and linewidth. Thus, the spectral distribution of atmospheric backscattering is composed of a narrow peak near the laser central frequency due to aerosol scattering and a much wider distribution from molecular scattering [11]. Filtering Mie scattering requires locking the central frequency of a narrow-bandwidth spectral filter and ultra-narrow-linewidth laser. The HSRL principle was described and demonstrated in 1968 by Fiocco and DeWolf [14], who conducted a non-ranging laboratory demonstration of separating aerosol and molecular backscattering. The HSRL technique was formalized in the early 1980s by the University of Wisconsin lidar group [15,16], who enabled the measurement of extinction cross-sections and aerosol-to-molecular backscatter ratios based on temperature profiles. In several subsequent works, the UW lidar group demonstrated the potential of HSRL to obtain profiles of various aerosols’ [17] and cirrus clouds’ [18,19,20] optical properties without making assumptions. Following studies indicated that HSRL is also useful for determining atmospheric temperature and pressure [21]. To increase HSRL sensitivity and reliability, the Colorado State University lidar group proposed using atomic vapor cells as blocking filters due to their higher spectral discrimination ratio, stability, and simplicity of instrumentation [22,23]. She et al. [24] retrieved the first vertical profiles of atmospheric state properties and aerosol radiative properties using a barium absorption filter. Later, Piironen and Eloranta [25] introduced the use of iodine cells, which improved HSRL performance and accuracy by allowing stronger absorption with a shorter cell at room temperature. Ever since, several new and diverse atmospheric applications have been reported, such as stratospheric temperature retrieval [26,27], wind-velocity retrieval [26,28], and aerosol classification [29,30]. HSRL has also been a valuable tool in other fields, such as ocean sciences, for estimating the ocean surface wind speed [31], investigating the marine atmospheric boundary layer [32], and retrieving the seawater volume scattering function [33] and depth-resolved optical properties [34]. Technical improvements have allowed lower costs and more durable and accurate systems. These include using different laser sources like multimode [35,36,37] and diode [38] lasers, in addition to different spectral filters like rubidium cells [38,39], and Fizeau [27] and Mach–Zehnder interferometers [35,40].

However, the HSRL technique has primarily been developed for long-range studies. Existing long-range HSRLs are unable to perform measurements close to the emission source due to two limitations: they are often blind in the first few meters because of the overlap problem, and their spatiotemporal resolution is inadequate. This is mainly because they emphasize the total measurement range by configuring the system to have a longer full-overlap distance, and spectral performance and signal accuracy by using low-repetition-rate, ultra-narrowband (and therefore long-pulse) lasers. To the best of our knowledge, only one HSRL designed for short-range operation has been reported. This is a CW near-infrared high spectral resolution Scheimpflug lidar (HSR-SLidar) for atmospheric aerosol sensing, developed by Chang et al. [39]. Based on the Scheimpflug principle, this system transmits an intensity-modulated 780 nm CW laser beam into the atmosphere. Then, it obtains the range-resolved atmospheric lidar signal from the pixel intensities of a tilted image sensor that satisfies the Scheimpflug principle. To match the laser wavelength, a temperature-stabilized rubidium cell was utilized as the spectral filter with a spectral discrimination ratio up to 10⁴. Due to this operational principle, the HSR-SLidar presents several inherent advantages, including its low cost, low instrumental complexity, and a blind overlap zone of less than 30 m. However, there are also some inherent disadvantages. First, the spatial resolution is range-dependent, decreasing with increasing measurement distance. Second, the pixel-distance relationship must be calibrated by measuring the beam spot from a distant building with a known distance. Additionally, it does not allow continuous, synchronized data acquisition, as it requires precise, step-wise wavelength switching of the CW laser. This causes instabilities in the laser wavelength, which require a dual-exposure scheme to solve. The first exposure is for waiting for wavelength stabilization, and the second one for performing lidar measurements. This process introduces extended integration periods, resulting in lower temporal resolution to ensure measurement fidelity.

To close the gap in highly accurate, high spatiotemporally resolved aerosol characterization at short ranges, this work presents the proof-of-concept of a short-range high spectral resolution lidar designed to measure within the first kilometer of the atmosphere, the blind zone of most long-range elastic and high spectral resolution lidars. To perform short-range measurements, the optomechanical alignment was configured to have a full overlap of the receiving optics’ field of view and the laser beam between 50 and 1000 m. To obtain a high spatial resolution and a correct suppression of the Mie backscattering component in the molecular channel, we used a commercial fiber laser with a linewidth of 300 MHz and a temporal pulse length of ∼5 ns. This laser is also wavelength-tunable, compact, and has a high repetition rate. As a blocking filter, we used a 40 cm long iodine cell to enhance the spectral performance by compensating the wide Mie backscattering spectra (due to the large laser linewidth) with its high aerosol scattering rejection ratio. We demonstrate that, despite its significantly large linewidth, the laser–iodine cell configuration exhibits satisfactory spectral performance, enabling discrimination between Mie and Rayleigh–Brillouin scattering spectra and the acquisition of calibrated aerosol backscatter coefficient profiles. Moreover, being compact in shape and lightweight, a future version of the proof-of-concept SR-HSRL presented in this work is suitable for prospective airborne lidar measurements. In Appendix A,

Table A1. HSRL systems for aerosol studies developed in the last 25 years.

HSRL Characteristics
Author Year	Laser						Spectral Filter			Resolution
	Type	$\mathsf{\lambda }$ (nm)	frep (Hz)	E (mJ)	$\mathsf{\tau }$ (ns)	$\mathsf{\Delta }\mathsf{\upsilon }$ (MHz)	Type	Temp. or Press.	FSR (GHz)	Temporal (min)	Spatial (m)
Hoyos-R.	L10	532	50 $\times {10}^3$ to 65 $\times {10}^3$	1 $\times {10}^{-3}$ to 1 $\times {10}^{-2}$	4.96	300	F1	40 °C	1.8	0.1	7.5
Chang 2025 [39]	L6	780	–	2 W	–	1	F8	60 °C 1.2 mPa	–	1	0.2–80
Gao 2023 [40]	L2	532	10	150	–	–	F6	–	0.3	0.5	1.5
Jin 2022 [56]	L3	355 532	10	100	4–6	173 116	F3	–	2.4	15	90
Wang 2022 [37]	L4	532	20	1.5	15	–	F4	–	3.0	5	15
Ke 2022 [60]	–	532	30	21	15	<10	F1	39 °C	–	–	24
Xu 2021 [27]	L5	532	100	120	–	–	F7	–	8.4	5	300
Wang 2021 [61]	L3	532	10	100	10	–	F1	–	–	2	75
Jin 2020 [13]	L3	355	10	100	4–6	173	F3	–	2.4	15	90
Liu 2019 [44]	–	532	40	150	–	–	F1	39 °C	–	–	480
Sannino 2019 [52]	L1	532	20	100	–	–	F9	–	–	–	1000
Razenkov 2018 [54]	L9	532	4 $\times {10}^3$	5 $\times {10}^{-2}$	2.7 $\times {10}^5$	<50	F1	–	1.8	–	–
Shen 2018 [51]	L1	355	50	350	–	90	F5	–	8	5	30
Burton 2018 [62]	L1	355	200	11	–	–	F4	–	2.0	12	15
		532					F1	65 °C	2.0
Jin 2017 [35]	L2	532	10	200	4–6	116	F2	–	0.255	15	1.5
Hayman 2017 [38]	L6	780	7 $\times {10}^3$	5 $\times {10}^{-3}$	1 $\times {10}^3$	77	F8	65 °C	–	1	150
Liu 2016 [55]	L1	532	–	270	–	100	F4	–	3.0	–	7.5
Bruneau 2015 [28]	L1	355	20	50	7	70	F10	–	1.5	–	60
Zhao 2015 [49]	L1	532	30	150	8	<90	F1	60 °C	2.0	5	7.5
Hoffman 2012 [58]	L1	532	10	160	6	<150	F9	–	7.5	1	45
Wirth 2009 [57]	L8	532	100	100	7.5	<150	F1	53–159 Pa	2.0	∼0.18	15
Liu 2009 [63]	–	532	2800	1.79	20	99% SP	F1	65 °C	–	45	100
Esselborn 2008 [64]	L7	532	100	110	–	–	F1	53–159 Pa	2.0	∼0.18	15
Li 2008 [32]	L1	532	10	80	10	100	F1	–	–	–	37.5
Hair 2008 [53]	L1	532	200	2.5	15	38	F1	65 °C	2.0	∼0.17	60
Liu 2006 [50]	L1	354.7	20	200	–	–	F5	–	2.5	4	30
Hair 2001 [12]	L1	532	20	300	5	74	F1	57 °C 82 °C	3.0 4.3	3	75
Liu 1999 [48]	L1	532	10	400	5–7	90	F1	–	–	15	>7.5

compares the spatiotemporal resolution and laser and spectral filter specifications of the SR-HSRL developed in this work (first row) to the HSRLs developed in the last 25 years for aerosol characterization.

This paper is structured as follows. Section 2 presents the HSRL framework, inversion method, and lidar architecture. Then, Section 3 illustrates the campaign overview and the experimental results. Afterwards, Section 4 discusses the advantages of the developed SR-HSRL in terms of retrieval range and spatiotemporal resolution compared to other systems reported in the literature. Finally, Section 5 presents the conclusions of this work and perspectives of the short-range HSRL we have developed.

2. Materials and Methods

2.1. HSRL Framework

As previously mentioned, the spectral distribution of light backscattered by the atmosphere is composed of a narrow peak corresponding to the lower frequency shift of aerosol scattering and a wide distribution from molecular scattering, as can be seen in the lower-left corner of Figure 1. The measured signal is split into two parts: one is directed straight to the detector of the total channel, and the other passes through a spectral filter that cuts the Mie signal and allows the Rayleigh return to pass through to the molecular (or HSR) channel. By measuring two different signals for the two atmospheric variables at each height, the lidar problem turns into a well-posed problem that can be solved without any assumptions.

Dropping the wavelength dependence for simplicity, the signals $P_{hsr}\left(r\right)$ and $P_{tot}\left(r\right)$ measured by the HSR and total channels are described by Equations (1) and (2) [42], respectively.

$$\begin{array}{c}\hfill P_{hsr}\left(r\right)=K_{hsr}{\displaystyle \frac{O_{hsr}\left(r\right)}{r^2}}\left[X_m{\beta }_{mol}\left(r\right)+X_p{\beta }_{aer}\left(r\right)\right]exp\left[-2{\int }_0^r\left[{\alpha }_{mol}\left(r^{\prime }\right)+{\alpha }_{aer}\left(r^{\prime }\right)\right]\mathrm{d}{r}^{\prime }\right]\end{array}$$

(1)

$$\begin{array}{c}\hfill P_{tot}\left(r\right)=K_{tot}{\displaystyle \frac{O_{tot}\left(r\right)}{r^2}}\left[{\beta }_{mol}\left(r\right)+{\beta }_{aer}\left(r\right)\right]exp\left[-2{\int }_0^r\left[{\alpha }_{mol}\left(r^{\prime }\right)+{\alpha }_{aer}\left(r^{\prime }\right)\right]\mathrm{d}{r}^{\prime }\right]\end{array}$$

(2)

where ${\displaystyle K_{hsr,tot}=P_{0}A\frac{c\tau }{2}{\eta }_{hsr,tot}}$ is the radiometric constant of each channel, given by the average power of the laser pulses $P_0$, the effective spatial pulse length $c\tau /2$, the area of the receiving telescope A, and the overall system efficiency ${\eta }_{hsr,tot}$. The geometric dependency is given by the overlap function of each channel $O_{hsr,tot}\left(r\right)$ and the quadratic decrease of the signal intensity with distance $r^{-2}$ as the light waves propagate throughout the atmosphere. Considering that the spectral filter is not ideal, and thus some parasitic particle signal will be transmitted through the molecular channel, its equation accounts for the effective transmittance of both Rayleigh–Brillouin and Mie scattering spectra, $X_m$ and $X_p$.

The molecular backscatter and extinction coefficients ${\beta }_{mol}\left(r\right)$ and ${\alpha }_{mol}\left(r\right)$ can be estimated using Rayleigh’s scattering theory and additional atmospheric temperature and pressure profiles, as in Equations (3) and (4) [42].

$$\begin{array}{cc}\hfill {\alpha }_{mol}\left(r\right)& =N_m\left(r\right)\sigma \hfill \end{array}$$

(3)

$$\begin{array}{cc}\hfill {\beta }_{mol}\left(r\right)& ={\displaystyle \frac{{\alpha }_{mol}\left(r\right)}{8\pi /3}}\hfill \end{array}$$

(4)

The Rayleigh extinction cross-section $\sigma$ and the range-dependent number density of molecules $N_{mol}\left(r\right)$, which depend on the temperature, pressure, and CO₂ content of the atmosphere, are defined as in Equations (5) and (6), respectively [43].

$$\begin{array}{cc}\hfill & \sigma ={\displaystyle \frac{24{\pi }^3{\left(n_s^2-1\right)}^2}{{\lambda }^4{N}_s^2{\left(n_s^2+2\right)}^2}}{\displaystyle \frac{6+3\gamma }{6-7\gamma }}\hfill \end{array}$$

(5)

$$\begin{array}{cc}\hfill & N_{mol}\left(r\right)=N_s{\displaystyle \frac{T_s}{P_s}}{\displaystyle \frac{P\left(r\right)}{T\left(r\right)}}\hfill \end{array}$$

(6)

where $n_s$ is the refractive index for standard air at the given wavelength, $N_s$ is the molecular number density for standard air, $T_s$ and $P_s$ are the temperature and pressure of standard air, and $\gamma$ is the wavelength-dependent depolarization factor that describes the effect of molecular anisotropy [43].

2.2. Inversion Method

The first step is to pre-process the data by performing background removal and range correction. Afterward, we calculate the interchannel calibration constant C to estimate the corrected total channel $P_{tot,corr}$ in the framework of the HSR channel. To this end, we measured out of the I₂ absorption band so that most of the aerosol and molecular frequencies are transmitted in the HSR channel (meaning $X_m\approx X_p\approx 1$). Thus, the signals should be equal in both channels, the only difference being the overlap functions and radiometric constants. By doing the ratio of the offline HSR to the total channel, we estimated the interchannel calibration constant as shown in Equation (7).

$$C\left(r\right)={\displaystyle \frac{P_{hsr}\left(r,\nu ={\nu }_{\mathrm{off}}\right)}{P_{tot}\left(r\right)}}={\displaystyle \frac{K_{hsr}{O}_{hsr}\left(r\right)}{K_{tot}{O}_{tot}\left(r\right)}}$$

(7)

Then, the corrected total channel power can be retrieved as

$$P_{tot,corr}\left(r\right)=C{P}_{tot}\left(r\right)$$

(8)

After estimating C, we need to calculate the Mie crosstalk parameter ${\chi }_p$, i.e., how much the spectral filter transmits parasite aerosol signals. To this end, we utilize cloud or aerosol zones measured by the HSR channel, which, in an ideal case where no Mie signal is transmitted, should not be detected. We optimized the value of ${\chi }_p$ until the Mie signals were removed. Thus, the corrected molecular or Rayleigh channel $P_{Ray}$ is defined as

$$P_{Ray}\left(r\right)=P_{hsr}\left(r\right)-{\chi }_p{P}_{tot,corr}\left(r\right)$$

(9)

Let us introduce the backscatter ratio (BR) defined as the ratio of the total backscatter coefficient to the molecular backscatter coefficient,

$$BR\left(r\right)={\displaystyle \frac{{\beta }_{aer}\left(r\right)+{\beta }_{mol}\left(r\right)}{{\beta }_{mol}\left(r\right)}}$$

(10)

The aerosol backscatter coefficient ${\beta }_{aer}$ can then be estimated from the backscatter ratio and the molecular backscatter coefficient ${\beta }_{mol}$, itself calculated from radiosonde data using Equation (4), with the following equation:

$${\beta }_{aer}\left(r\right)=\left[BR\left(r\right)-1\right]{\beta }_{mol}\left(r\right)$$

(11)

For HSRL retrieval, in order to compute BR, it is necessary to calculate the attenuated backscatter profiles $U_{Ray}$ and $U_{tot}$ using the radiometric constants $K_{Ray,tot}$ and the overlap functions $O_{Ray,tot}\left(r\right)$ as shown in Equations (12) and (13). It is essential to emphasize that, unlike long-range lidar systems, the overlap function cannot be ignored (assumed to be 1) in the short range since it is in this zone—typically within the first 1000 m, depending on the system configuration—where there is a variation from 0 to 1. Failing to account for the geometric factor in this area results in signal loss due to incomplete overlap.

$$\begin{array}{c}\hfill \begin{array}{cc}\hfill U_{Ray}\left(r\right)& =\left[X_m{\beta }_{mol}\left(r\right)+X_p{\beta }_{aer}\left(r\right)\right]exp\left[-2{\int }_0^r\left[{\alpha }_{mol}\left(r^{\prime }\right)+{\alpha }_{aer}\left(r^{\prime }\right)\right]\mathrm{d}{r}^{\prime }\right]\hfill \\ \hfill & ={\displaystyle \frac{P_{Ray}\left(r\right)r^2}{K_{Ray}{O}_{Ray}\left(r\right)}}\hfill \end{array}\end{array}$$

(12)

$$\begin{array}{c}\hfill \begin{array}{cc}\hfill U_{tot}\left(r\right)& =\left[{\beta }_{mol}\left(r\right)+{\beta }_{aer}\left(r\right)\right]exp\left[-2{\int }_0^r\left[{\alpha }_{mol}\left(r^{\prime }\right)+{\alpha }_{aer}\left(r^{\prime }\right)\right]\mathrm{d}{r}^{\prime }\right]\hfill \\ \hfill & ={\displaystyle \frac{P_{tot}\left(r\right)r^2}{K_{tot}{O}_{tot}\left(r\right)}}\hfill \end{array}\end{array}$$

(13)

Solving Equations (1) and (2), the BR is calculated from the ratio of $U_{Ray}$ and $U_{tot}$, and the molecular and particle backscattering transmittance through the iodine cell $X_m$ and $X_p$, as described in Equation (14) [44].

$$BR\left(r\right)={\displaystyle \frac{\left(X_m-X_p\right)U_{tot}/U_{Ray}}{1-X_p{U}_{tot}/U_{Ray}}}$$

(14)

2.3. Lidar Design

The proof-of-concept lidar described in Figure 2 is a short-range high spectral resolution lidar (SR-HSRL) designed as a collaboration between ONERA—the French Aerospace Lab—and NIES, the National Institute of Environmental Studies of Japan. It is designed to measure the aerosol Mie and molecular Rayleigh backscatter signals separately in the low PBL with high range and time resolution.

Table 1. NIES-ONERA SR-HSRL lidar specifications.

Laser	Wavelength	532.11357 to 532.20615 nm
	Wavelength stability	<0.1 pm
	Linewidth	<400 MHz
	Linewidth stability	<400 MHz
	Pulse duration	4.96 ns
	Pulse repetition rate	50 to 65 kHz
	Pulse energy	0.90 to 11.30 $\mathsf{\mu }$J
	Divergence angle	1.5 mrad
	Beam diameter (1/e)	1 mm
	Beam quality	M2 = 1.3
	Dimensions	270 × 270 × 40 mm
	Weight	2.9 kg
Spectral filter	Type	Iodine cell
	Diameter	2 in
	Length	400 mm
	Absorption band wavenumber	18,809.82 cm−1
	Absorption band FWHM	1.8 GHz
	Cell temperature	40 °C
Receiver	Type	Schmidt–Cassegrain
	Diameter	203.2 mm
	Focal length	2032 mm
	F-number	10
Sensors	Type	Hamamatsu MPPC
	Reference	C13366
	Bandwidth	5 MHz
	Active area	9 mm2

presents the complete characteristics of the SR-HSRL.

The emission part of the instrument is crucial for the HSRL technique. Conventional HSRL instruments utilize injection-seeded Q-switched Nd:YAG laser sources with high-energy pulses (several mJ) and low repetition rates (several tens of Hz). In this work, we used an amplified fiber laser operating at 532 nm with low-energy pulses (∼8 $\mathsf{\mu }$J) and a high repetition rate (50 to 65 kHz). The laser is used in a bi-static multi-axial configuration to enable lower-altitude measurements in the first-kilometer range by controlling the overlap range, inspired by previous short-range EBL works [6]. In the receiver part, a 10:90 beam splitter reflects 10% of the signal to the total channel and transmits 90% of it to the HSR channel, which uses a 40 cm long iodine (${\mathrm{I}}_2$) cell to filter the aerosol component of the backscattered light. The laser wavelength is matched to the minimum transmittance band of the iodine filter by temperature tuning to block the aerosol Mie component. Then, two thermoelectrically cooled SiPM Multi-Pixel Photon Counter (MPPC) sensors are used in both channels with a fast analog-to-digital converter (up to 160 MHz). With this configuration, the achievable signal-to-noise ratio (SNR) is 203 for the HSR channel and 471 for the total channel for the 6 s temporal resolution.

2.4. Laser Source

One of the novelties of this work is that the HSRL utilizes a commercial fiber laser source that is compact, has a high repetition rate, and enables short-range measurements due to its short temporal pulse length. The repetition rate can be tuned from 50 to 65 kHz, and the average power from 45 to 565 mW; thus, the energy varies from 0.69 to 11.30 $\mathsf{\mu }$J. When operating at minimum power (45 mW), power stability is 3.53% over a 16 h measurement. However, increasing the power to the maximum (565 mW) improves the stability (1.65%). The laser housing is 270 × 270 × 40 mm and weighs only 2.9 kg, making it ideal for integration on a portable HSRL system. Table 1 presents the complete laser specifications.

As mentioned earlier, performing short-range HSRL entails finding a compromise between the laser temporal pulse length, linewidth, and spectral and power stability. Additionally, the prospective airborne lidar applications of our work required a compact and lightweight laser. Finding a commercial laser fulfilling all these requirements was not possible, so Keopsys (a Lumibird company) customized its PGFL model [45] for our work. As a result, our laser source has a pulse duration of 4.96 ns, which corresponds to a theoretical spatial resolution $\mathsf{\Delta }R<80$ cm, making it ideal for performing short-range measurements. However, it must be considered that the bandwidth and frequency of the sensors and the signal-acquisition system will define the actual resolution. On the other hand, a Fourier-limited Gaussian laser with such a pulse length would have a linewidth >90 MHz. In reality, the laser linewidth is by design 300 MHz, so we could expect a spectral discrimination ratio of ∼300 for the iodine cell used. Most HSRL systems employ lasers with a much narrower linewidth, achieving higher spectral performances at the expense of spatial resolution. We have chosen this laser source because, for our application, we have a satisfactory compromise between both characteristics. In spectral terms, stability is as crucial as a narrow linewidth for the performance of the lidar system. We measured the wavelength and linewidth stability using a wavelength meter (HighFinesse WS/6-200 VIS/IR-II) with a resolution of 100 MHz in wavelength and 400 MHz in linewidth. We measured the spectral stability for 17.5 h, finding that the linewidth has a constant value, limited by the instrument’s resolution, of <400 MHz, and that the wavelength has a stability <100 MHz. Both results, displayed in Figure 3, are satisfactory for performing HSRL measurements.

Another interesting feature of this laser is that it can be wavelength-tuned, which makes it suitable for use with an iodine cell. This is achieved by changing the temperature in steps of 0.01 °C, which results in a wavelength increase of 37 pm. After characterizing the wavelength-tuning process, the wavelength in nm can be determined for any temperature value using the equation $\lambda \left(T\right)=0.037T+531.41$. Moreover, we found that the wavelength tuning is linear (R² = 1), which ensures repeatability when scanning the iodine spectral bands to find the ideal transmittance–full-width half-maximum (FWHM) combination to optimize the cell performance, i.e., maximizing the Rayleigh signal’s transmittance while minimizing the Mie component. Several scanning tests were performed, finding twelve iodine absorption bands (1187 to 1198) in the laser wavelength range (532.11357 to 532.20615 nm), with transmittances of 10⁻²to 10⁻³ and FWHMs between 0.6 and 3.1 GHz, with most lines being around 1.5 GHz, depending on the cell operating temperature. After the scanning tests, we selected the iodine line at wavenumber 18,809.82 cm⁻¹ (laser wavelength of 532.197 nm) because it has a combination of transmission and FWHM that maximizes Mie scattering suppression and Rayleigh–Brillouin transmittance.

3. Results

3.1. Proof-of-Concept Campaign Overview

The prototype of SR-HSRL was optimized to measure the first thousand meters of the atmosphere, retrieving aerosol backscatter coefficients from below 100 m. The spatiotemporal resolution, limited by the acquisition system, is 7.5 m and 6 s. Testing the prototype did not require a finer spatiotemporal resolution, as the primary goal was to demonstrate that our compact and high-repetition-rate fiber laser, despite having a linewidth of 300 MHz, has a satisfactory spectral performance to suppress Mie scattering in the HSR channel and is suitable for a future implementation of an SR-HSRL with a higher spatiotemporal resolution. A two-day measurement campaign was performed at NIES in Tsukuba, Japan, in July 2024 to test the prototype. Figure 4, where the gray color represents invalid values, shows the temporal evolution of the raw RCS signals for the part of the campaign from 11 July at 18:00 to 12 July at 05:00. The top panel corresponds to the HSR channel, and the bottom panel to the total channel. In the HSR channel, it can be observed that the aerosol component is correctly suppressed from the molecular signal for aerosol-only layers, although some crosstalk remains from cloud backscattering. The stripe-like variations in the HSRL signal seen in the upper-right region of the top panel are likely due to small and periodic fluctuations in the laser wavelength due to temperature-control issues in the seed laser. This wavelength drift leads to a mismatch with the iodine cell absorption band, resulting in differences in the transmission of the Mie scattering spectrum, and consequently in the stripe pattern observed. On the other hand, the total channel shows several aerosol layers from 150 m to 900–1500 m, with the height of the layers increasing at night (from 23:00), and some low-altitude clouds.

3.2. Data Pre-Processing

The first step for correcting the data is to find the interchannel calibration constant C, which encompasses the ratio of both channels’ overlap and radiometric constants, as described in Equation (7). Then, the total channel can be corrected by multiplying the measured signal by this constant, as in Equation (8). Afterwards, the correction of the HSR signals implies reducing the Mie crosstalk, for which we need to determine and optimize the crosstalk parameter ${\chi }_p$. Mie crosstalk was estimated and removed from 2 min averaged profiles. For each averaged HSR signal, a linear regression was applied in a cloud or high-intensity aerosol region against the expected molecular signal, which is represented as a straight line. Several ${\chi }_p$ values are tested for each averaged profile in an optimization algorithm, choosing the one that maximizes the coefficient of determination R² of the fitted curve. The crosstalk parameter values range from 2 $\times {10}^{-3}$ to 8.9 $\times {10}^{-2}$, with most of the averaged profiles having 8.9% Mie crosstalk.

To show the signals’ correction, Figure 5 compares some averaged profiles of uncorrected (darker lines) and corrected (lighter lines) signals. The profiles on top represent the molecular channel signals, while the total channel signals are displayed in red in the bottom panels. The left-most panels (21:00 to 21:30) show a low cloud from 300 to 400 m measured by both channels. The cloud peak, visible in the darker blue profile, illustrates Mie crosstalk in the HSR signal. This parasite signal is significantly suppressed after performing the crosstalk correction, as shown in the Rayleigh profile (lighter blue line). The middle panels (23:30 to 00:00) display some aerosol layers up to 1000 m, with no clouds. Here, the crosstalk correction in the Rayleigh channel enables the acquisition of a signal close to the reference molecular value. Lastly, from 04:30 to 05:00, two well-defined aerosol layers and a cloud (between 800 and 1000 m) are noticeable in the uncorrected channels. After the crosstalk correction, only a sparse cloud signal remains in the Rayleigh profile. As seen in all three timeframes, although the correction significantly reduces Mie crosstalk for strong-intensity aerosol layers or clouds, some undesired aerosol signals remain since over-correction results in unfeasible (negative) values in the Rayleigh signal. Having little to no crosstalk in the HSR channel would imply improving the iodine cell’s absorption by increasing its temperature or length. However, that would imply decreasing the transmission of the molecular signals, therefore reducing the system’s spectral performance.

After determining the crosstalk parameters and interchannel calibration constant, the signals were corrected for the chosen measurement period. The corrected RCSs are displayed for both channels in Figure 6, where the top panel corresponds to the Rayleigh channel (RCS_Ray) and the bottom one to the corrected total channel (RCS_{tot, corr}). It can be seen in the corrected total channel that the calibration constant C allows for a cleaner signal, i.e., the aerosol layers are more distinguishable. Moreover, it reduces the overlap height from 150 to less than 50 m, thereby increasing the ability to measure aerosols near the ground. The Rayleigh channel indicates that the Mie crosstalk has been substantially reduced, as the intensity of the cloud signals is significantly decreased. However, as aforementioned, completely removing Equation (8) would render the molecular signals unfeasible.

3.3. Determination of the Attenuated Backscatter Signals

To estimate the attenuated backscatter signals, and thus the aerosol backscatter coefficient, it is necessary to determine the radiometric constant and the overlap function for each channel. For the Rayleigh channel, where we assume a mostly pure molecular signal, a Rayleigh fit enables the determination of the radiometric constant. To this aim, we optimized the value of $K_{Ray}$ so that the measured signal fits the molecular RCS profile analytically calculated from the radiosonde data measured at midnight on 12 July 2024, at the Tateno meteorological station (36°03′29.2″N 140°07′32.9″E), located 1.4 km away from the measurement site (NIES facilities). In this case, the optimum value of K for this channel is 7 × 10¹⁰ Vm³sr. After determining $K_{Ray}$, we used Equation (1) to determine $O_{Ray}$ in a zone with negligible aerosols in the Rayleigh signal, i.e., a mostly pure molecular signal. Thus, it is safe to assume that ${\beta }_{aer}\left(r\right)={\alpha }_{aer}\left(r\right)=0$ and the geometric factor is given by Equation (15), where ${\beta }_{mol}\left(r\right)$ and ${\alpha }_{mol}\left(r\right)$ are calculated from the radiosonde data, and $X_m=5.9\times {10}^{-1}$, the effective transmittance of the Rayleigh–Brillouin scattering spectral throughout the iodine cell, was calculated from the cell’s characterization measurements.

$$O_{Ray}\left(r\right)={\displaystyle \frac{P_{Ray}\left(r\right)r^2}{K_{Ray}{X}_m{\beta }_{mol}\left(r\right)exp\left[{\displaystyle -2{\int }_0^r{\alpha }_{mol}\left(r^{\prime }\right)\mathrm{d}{r}^{\prime }}\right]}}$$

(15)

The obtained overlap function is displayed in Figure 7. After determining $K_{Ray}$ and $O_{Ray}\left(r\right)$, we use the interchannel calibration profile C, defined in Equation (7), to find the corresponding quantities for the total channel. In a zone of full overlap (from 550 m, as seen in Figure 7), the total channel radiometric constant was calculated as $K_{tot}=K_{Ray}/C=1.55\times {10}^{11}$ Vm³sr, where C has a mean value of 0.45 in the full overlap zone. Finally, from $K_{Ray}$, $K_{tot}$, and $O_{Ray}\left(r\right)$, we once again use Equation (7) to determine the overlap function of the total channel, which is shown in Figure 7.

The overlap function is a critical parameter that must be addressed for short-range measurements where the full-overlap assumption cannot be made. As previously shown [46,47], even small misalignments between the laser beam path and the receiver optical axis can significantly affect the overlap function at short ranges. The overlap function differs in both channels due to the different optical paths that arise from different optical elements. Firstly, the 90:10 beam splitter cube that splits the received atmospheric backscattering into the two channels can slightly modify the beam profile and divergence, leading to channel-dependent overlap. Moreover, the narrowband, 40 cm long iodine cell used in the molecular channel introduces additional differences in the optical path, alignment, and field of view of this channel regarding the total channel, also affecting the overlap function. Furthermore, the optical axes of the detectors in each channel are not identically positioned relative to the laser beam. Finally, coatings on the spectral and interference filters may preferentially transmit different angular components of the incoming light, again modifying the near-field overlap.

3.4. Estimation of the Aerosol Backscatter Coefficients

After correcting both signals, we used Equation (14) to calculate the backscatter ratio, where the transmittance of the molecular and particle backscattering spectra through the iodine cell $X_{m,p}$ were estimated from laboratory characterization measurements, obtaining $X_m=5.9\times {10}^{-1}$ and $X_p=8.83\times {10}^{-3}$. Afterward, the aerosol backscatter coefficient was straightforwardly calculated using Equation (11). For this, the molecular backscatter coefficient ${\beta }_{mol}$ was calculated from the corresponding radiosonde data. The temporal evolution of the estimated ${\beta }_{aer}^{HSRL}$ is shown in Figure 8, where the effective retrieval range is 50 to 1000 m.

It can be seen in the figure that, from 18:00 to 18:30, there is a thick aerosol layer from 300 to 1000 m with ${\beta }_{aer}^{HSRL}$ values ranging from $1.07\times {10}^{-6}$ to $8.06\times {10}^{-5}$ m⁻¹sr⁻¹. Then, from 18:30 to 19:00, the aerosol load is too low below 250 m and above 900 m, with a strong layer ranging from $1.01\times {10}^{-6}$ to $1.28\times {10}^{-4}$ m⁻¹sr⁻¹ at 440 m. In the next half hour, ${\beta }_{aer}^{HSRL}$ values are significant above 150 m but decrease in intensity, varying from $1.32\times {10}^{-6}$ to $8.77\times {10}^{-6}$ m⁻¹sr⁻¹. Up to 21:00, the aerosol content is rather low ($1.02\times {10}^{-6}$ to $2.04\times {10}^{-6}$ m⁻¹sr⁻¹), with thin, separate layers from 100 to 200 m, 250 to 400 m, and 450 to 550 m and a slightly more intense layer from 700 to 1000 m ($3.50\times {10}^{-6}$ m⁻¹sr⁻¹). From 21:00 to 21:30, there is a significant increase in the aerosol content, with a layer from 100 to 450 m and a peak intensity of $1.77\times {10}^{-4}$ m⁻¹sr⁻¹, and another one from 450 to 650 m ($2.43\times {10}^{-6}$ m⁻¹sr⁻¹). From 21:30 to 22:30, there is a cloud from 300 to 500 m, with backscattering coefficient values as high as $2.31\times {10}^{-3}$ m⁻¹sr⁻¹. For the next hour and a half, there is a medium-intensity aerosol layer from 50 m and centered at 400 m whose maximum value is $3.05\times {10}^{-5}$ m⁻¹sr⁻¹, with a sparse layer ($5.61\times {10}^{-6}$ m⁻¹sr⁻¹) between 700 and 900 m in the end. Then, from midnight to 01:30, the aerosol load increases in height and intensity, with multiple layers extending up to 950 m, especially the high-intensity layer from 700 to 800 m (00:30 to 01:00), where the maximum ${\beta }_{aer}^{HSRL}$ value is $2.98\times {10}^{-4}$ m⁻¹sr⁻¹ at 750 m. Then, from 01:30 to 02:30, there is a cloud from 300 to 500 m, where the backscattering is as high as $1.96\times {10}^{-3}$ m⁻¹sr⁻¹ at 450 m. It is worth noting that, unlike EBL lidar, the HSRL method can retrieve backscattering values inside and above clouds, as can be seen around 02:00, where there is a layer from 550 to 1000 m with peaks at 832 m ($1.09\times {10}^{-4}$ m⁻¹sr⁻¹) and 968 m ($5.16\times {10}^{-4}$ m⁻¹sr⁻¹). After this, the aerosol load separates into two distinct thick layers. The bottom one has a maximum backscattering intensity of approximately $4.19\times {10}^{-6}$ m⁻¹ sr⁻¹. By contrast, there is a cloud in the upper one, reaching values up to $2.75\times {10}^{-3}$ m⁻¹sr⁻¹.

To accurately interpret large contrasts in backscattering, it is crucial to consider the role of measurement uncertainties. Error propagation in lidar data is an important aspect that requires attention for accurate analysis; however, this study does not comprehensively address it. We performed a Monte Carlo error analysis to evaluate the impact of introducing noise levels to the main variables involved in the calculation of ${\beta }_{\mathrm{aer}}\left(r\right)$ (see Equation (11)). The parameters we measured include the particle and molecular transmission through the iodine cell, i.e., $X_p$ and $X_m$, and ${\beta }_{\mathrm{mol}}\left(r\right)$, whose error is related to atmospheric temperature variations. Our preliminary results indicate that error magnitudes tend to increase during significant aerosol or cloud events. As a result, we focused our analysis on aerosol occurrences. Based on laboratory characterization measurements, we considered a relative error of 1.0% and 6.5% for the molecular transmission $X_m$ and particle transmission $X_p$, respectively. Additionally, we introduced a 3 K relative error in atmospheric temperature to assess its impact on molecular backscattering [37], finding a 1% variation in ${\beta }_{\mathrm{mol}}\left(r\right)$. As a result of the MC method, we found that these three variables can induce errors up to 7.2% in ${\beta }_{\mathrm{aer}}\left(r\right)$. This result emphasizes the need for a dedicated sensitivity analysis to quantify the influence of each variable on the uncertainty of the aerosol retrievals.

4. Discussion

The major novelty of the short-range HSRL presented in this work is its ability to perform lidar measurements and retrieve quantitative aerosol backscattering coefficients in the PBL (within the first thousand meters) with high spatiotemporal resolution, using a high-repetition-rate diode-based commercial laser and an iodine cell. From an instrumental point of view, the system is compact and cost-effective. Unlike conventional long-range HSRLs, which are based on high-energy, low-repetition-rate injection-seeded Q-switched Nd:YAG lasers, our system utilizes an amplified fiber laser at 532 nm with low-energy pulses (∼8 $\mathsf{\mu }$J) and a high repetition rate (50 to 65 kHz). Doing HSRL in the short range with a high spatial resolution implies making a compromise between the laser pulse length (4.96 ns), linewidth (300 MHz), and full overlap distance (50 to 1000 m, defined by the optomechanical alignment of the receiving optics). The long-range atmospheric HSRLs reported in the literature have a clear priority for spectral width over pulse duration, as their primary focus is on spectral discrimination rather than spatial resolution. According to E. Eloranta [11], HSRL requires an ultra-narrow-width laser (<100 MHz) for performing accurate spectral discrimination. Most HSRLs reported following this guideline [12,28,38,48,49,50,51,52,53,54], with values ranging from 38 MHz (15 ns) for NASA’s airborne HSRL-2 [53] to 100 MHz (10 ns) for Chinese shipborne [32] and ground-based [55] systems. Other authors report linewidth values slightly larger; the three HSRLs developed by Jin et al. [13,35,56] have a linewidth of 116 MHz at 532 nm or 173 MHz at 355 nm (both 4–6 ns). DLR’s multilidar airborne system [57] performs HSRL measurements at 532 nm with light pulses <150 MHz (7.5 ns), as Hoffman et al. [58] (6 ns). In this work, we demonstrate that it is possible to perform HSRL measurements with a wide-linewidth laser, achieving a spectral discrimination ratio of $300$, which is enough for detecting several different aerosol layers while keeping a high range (7.5 m) and temporal (6 s) resolution.

From a data-analysis perspective, due to the optimized configuration to have a low blind zone and the high spatial resolution, our SR-HSRL prototype is a feasible instrument for studying the lower atmosphere as illustrated in the quicklook of Figure 8 and as highlighted in a sample averaged profile in the right panel of Figure 9, where it can be seen that the aerosol backscatter values are retrieved from 50 to 1000 m. Another important advantage of the HSRL retrieval method is that it relies on the direct calculation of the backscatter ratio rather than on numerical integration methods such as the Klett or Fernald algorithms used in EBL. Since integral methods can diverge in the presence of strong extinction, they often fail above optically thick layers. In contrast, the HSRL approach remains stable, enabling retrievals within and above cloud layers, as long as the laser beam is not completely extinguished. Thus, HSRLs are suitable instruments for studying clouds and aerosol–cloud interaction processes. The left panel of Figure 9 demonstrates this capability, showing a cloud located from 300 to 550 m, with cloud backscattering reaching up to 10⁻¹ m⁻³sr⁻¹). In addition to the cloud, the measurement also captures an aerosol layer and another cloud positioned above.

As previously mentioned, SR-EBLs have proven to be highly effective tools for measuring particles near the source with high spatiotemporal resolution [5,6,7,8,59]. These systems have introduced new diagnostic capabilities for various aeronautics and environmental applications, such as measuring pollution emitted by airplane engines and characterizing the atmosphere in the lower PBL, where long-range lidars are blind. However, the inherently ill-posed EBL problem limits the accuracy of these instruments. Additionally, for ultra-short-range measurements, no molecular reference zone is available, so the inversion of the signals requires a dedicated forward inversion method without any knowledge of boundary conditions [7]. As a result, assessing ultrafine particles like freshly emitted soot poses a significant challenge for EBLs. The short-range HSRL developed in this work will enable closing this gap by characterizing coarse to ultrafine aerosols at short ranges with high accuracy and spatiotemporal resolution. This is possible because HSRL is a well-posed problem that can be solved without assumptions on the lidar ratio, allowing for a straightforward inversion of the particles’ radiometric and microphysical properties. With this enhanced accuracy, the SR-HSRL will be capable of measuring smaller particles while maintaining the short-range capabilities characteristic of the SR-EBL instruments.

Compared to the only short-range HSRL reported in the literature [39], which, as mentioned, is based on the Scheimpflug principle, the pulsed short-range HSRL presented in this work is instrumentally more complex, but has a similar advantage in terms of blind overlap zone height (30 m vs. 50 m). Moreover, it overcomes the other HSR-SLidar disadvantages. The spatial resolution is constant, there is no distance calibration since it is based on the time-of-flight principle, and it allows continuous, synchronized data acquisition as the laser does not undergo wavelength switching during measurements.

5. Conclusions

The feasibility of short-range implementation of the high spectral resolution lidar (SR-HSRL) technique was demonstrated through the development and evaluation of a compact SR-HSRL prototype, integrating an iodine absorption cell and a non-conventional, high-repetition-rate diode-based laser with a 300 MHz linewidth and 5 ns pulse duration. Developing this instrument presented a significant challenge, both scientifically and technically. One of the technological challenges was finding a spectrally fine and stable laser source to perform HSRL measurements with satisfactory spectral resolution and whose pulse width is in the order of a few nanoseconds to perform short-range measurements with adequate spatiotemporal resolution. This implied developing a methodology for the complete characterization of different lasers, as manufacturers do not provide the complete spectral and pulse features. Moreover, the laser had to be affordable, compact, tunable, and with a high repetition rate; the opposite of the lasers generally used for HSRL, except for the diode-laser-based HSRL developed by Hayman and Spuler [38], which, however, is a long-range lidar with a 150 m resolution.

We validated the instrument through a successful proof-of-concept campaign, retrieving aerosol backscatter coefficient profiles from 50 to 1000 m with a spatial resolution of 7.5 m and a temporal resolution of 6 s. The results demonstrate that the HSRL system is effective for investigating clouds and cloud–aerosol interactions, as it enables measurements both within and above cloud layers. In future work, we aim to develop an ultra-short-range version of the instrument to study diverse aerosol types near their sources, particularly within the first 100 m, using both ground-based and mobile platforms. This future work will focus on comprehensive characterization of aerosol radiative properties, including retrieval of aerosol extinction coefficients and lidar ratios in addition to backscatter coefficients. These advancements will enhance our understanding of aerosol behavior and its impact on atmospheric processes.