Analysis of Radiative Transfer Characteristics for Underwater Hyperspectral LiDAR

Highlights

What are the main findings?

• A wavelength-dependent radiative transfer model for underwater hyperspectral LiDAR (UDHSL) was established and quantitatively validated against independent AC-S spectrometer measurements, with fitted attenuation coefficients agreeing within +3.4% and Beer–Lambert linearity yielding R² > 0.99.
• Normalized echo intensity exhibits a strong linear relationship with target reflectance, validated by regressions using both nominal (R² = 0.938) and independently measured reflectance (R² = 0.873); per-band analysis yields mean R² = 0.915 (nominal) and 0.851 (measured) across 500–660 nm.
• Effective detection range decreases from >10 m in clear water (∼0.05 m⁻¹) to ∼5–6 m in turbid water (∼0.3 m⁻¹), accompanied by a shift in the optimal wavelength window from 450–550 nm to 530–570 nm.

What are the implications of the main findings?

• The quantitatively validated model and attenuation data provide an experimental basis for radiometric calibration and underwater target reflectance retrieval under controlled freshwater laboratory conditions.
• The characterized wavelength-dependent attenuation behavior offers first-order engineering guidance for selecting optimal working wavelengths and designing UDHSL systems for different water quality conditions.

Abstract

Targeting the long-term goal of synchronous acquisition of underwater terrain and material composition information, this study establishes a radiative transfer model for underwater hyperspectral LiDAR (UDHSL) and systematically verifies the effects of target reflectance, detection distance, and laser wavelength on backscattering echo intensity through controlled laboratory experiments. A wavelength-dependent water attenuation correction term incorporating absorption and scattering was introduced into the conventional LiDAR equation to derive a hyperspectral LiDAR radiative transfer equation applicable to underwater environments, and a normalized echo intensity processing method using window glass reflection as a reference was proposed. This study uses a custom-built UDHSL system (wavelength range: 450; detection range approximately 5–6 m). The echo intensity exhibits pronounced wavelength selectivity, peaking at 450–550 nm in clear water and shifting to 530–570 nm in turbid water. These experimental results are consistent with theoretical predictions of the radiative transfer model, validating its fundamental correctness and providing an experimental basis for radiometric calibration and underwater target reflectance retrieval of UDHSL systems.

1. Introduction

Underwater exploration is a fundamental endeavor for marine resource development and deep-sea scientific research, with the core requirement of synergistically acquiring and accurately interpreting seabed topographic morphology and material composition. Geodynamic studies have demonstrated significant spatial coupling relationships between seafloor topographic features and sediment types. By establishing multi-source data fusion models integrating terrain and material information, visual reconstruction of spatial distribution patterns of underwater targets can be achieved, providing multi-physical-field constraints for mineral resource exploration and submarine topographic mapping. Among conventional underwater detection methods, multibeam sonar enables wide-area terrain surveying but lacks material identification capability, while passive optical remote sensing is limited by insufficient illumination in deep-sea environments. Light detection and ranging (LiDAR) technology exploits the active detection characteristics for underwater applicability.

Hyperspectral LiDAR (HSL) leverages the multi-wavelength simultaneous detection capability of supercontinuum laser pulses to overcome the information dimensionality limitations of conventional single-wavelength LiDAR systems [1,2,3]. In recent years, various HSL systems have been developed, including full-waveform HSL [4], liquid crystal tunable filter (LCTF)-based HSL [5,6], and multi-channel HSL [7,8,9], which enable synchronous acquisition of target geometric morphology and spectral reflectance characteristics in a single measurement, forming integrated spatial–spectral datasets [10,11]. The backscattering intensity of HSL contains rich information regarding surface reflectance properties. Previous studies have confirmed significant correlations between HSL backscattering intensity and surface roughness [4], moisture content [10], and mineral composition [12], with notable application achievements in vegetation classification [13,14] and crop monitoring [15]. However, all of the aforementioned studies were conducted in air media. Extending HSL technology to underwater detection scenarios introduces new challenges: laser propagation in water is subject to dual attenuation from absorption and scattering, with the attenuation coefficient exhibiting significant wavelength dependence; multiple factors including detection distance, target reflectance, and system parameters jointly influence the backscattering echo intensity, making accurate extraction of the intrinsic target reflectance from echo signals a complex task.

Currently, studies on the radiative transfer characteristics of HSL have primarily focused on echo intensity correction in air media [16]. In the field of underwater LiDAR, oceanographic profiling LiDAR has been employed for seawater optical parameter measurement and water scattering characterization [17], with these efforts focusing primarily on single-wavelength systems. More recently, multispectral and full-waveform LiDAR techniques have been applied to underwater target identification and classification: Chen et al. [18] classified underwater mineral ores using a tunable-laser multispectral LiDAR, Du et al. [19] demonstrated sediment type classification based on airborne LiDAR full-waveform data with an overall accuracy of 94.1%, and Ma et al. [20] employed an AOTF-based multispectral LiDAR for underwater object classification. These studies primarily addressed classification algorithms and data-driven target identification. The present work, in contrast, focuses on radiative transfer modeling, seeking to quantitatively characterize the coupled effects of target reflectance, detection distance, water optical properties, and laser wavelength on echo intensity. Our research group has previously reported two related studies on the UDHSL system: a system-level study [21] that detailed the engineering design, system integration, ranging accuracy, and three-dimensional hyperspectral point cloud acquisition capabilities, and a preliminary conference paper [22] that briefly demonstrated the feasibility of acquiring underwater echo signals. However, neither study established a quantitative radiative transfer model or systematically characterized the relationships among echo intensity, target reflectance, detection distance, and laser wavelength. Systematic quantitative modeling and experimental verification of the radiative transfer characteristics of underwater HSL backscattering intensity—modulated by the combined effects of these factors—thus represent an open research direction that the present work seeks to address.

To address these gaps, the present study develops a UDHSL system (wavelength range: 450–700 nm; spectral resolution: 10 nm; maximum detection distance: >27 m). Building upon the conventional atmospheric LiDAR equation, a wavelength-dependent water attenuation correction term incorporating absorption and scattering is introduced to establish a hyperspectral LiDAR radiative transfer model applicable to underwater environments. A normalized intensity processing method based on window glass reflection is also proposed. Two sets of single-variable controlled experiments were designed and conducted: one using standard diffuse reflectance panels with 10 nominal reflectance values (20–90%) to verify the quantitative relationship between echo intensity and target reflectance, and another performing multi-distance detection in both clear-water pipe (attenuation coefficient ≈0.05 m⁻¹ at 550 nm) and turbid-water pool (attenuation coefficient ≈0.3 m⁻¹ at 550 nm) environments to investigate the attenuation behavior of echo intensity with respect to distance and wavelength.

2. Materials and Methods

2.1. UDHSL System Overview

The underwater hyperspectral LiDAR system utilizes a supercontinuum laser source [23,24] to generate broadband laser pulses. Wavelength selection is achieved through tunable optical filters, and the time-of-flight (ToF) method is employed for precise target ranging to acquire seabed topography. Simultaneously, the spectral reflectance information of seabed features is obtained through echo intensity acquisition.

Figure 1 illustrates the architecture of the underwater supercontinuum hyperspectral LiDAR system. The system comprises a supercontinuum laser, a tunable filter, a receiving optical subsystem, a detector, a high-speed digital acquisition module, a two-dimensional scanning module, a control module, and a waterproof shell. Broadband pulsed laser light generated by the supercontinuum laser passes through a tunable bandpass filter for dynamic wavelength selection, producing a quasi-monochromatic laser output with high spectral purity. The collimated beam is directed by a two-dimensional scanning module for wide-field-of-view spatial pointing control. After reflection from the detection target, the returned signal is efficiently collected by a large-aperture receiving optical system and focused onto a high-sensitivity detector. Full-waveform signal capture is realized using a high-speed digital acquisition module, enabling synchronous analysis of target characteristics in both temporal and spectral dimensions.

The system employs a multi-layer coordinated control architecture, precisely synchronizing laser emission, rapid spectral switching, and high-speed spatial scanning through a customized communication protocol. To withstand extreme underwater conditions, a composite-material waterproof shell integrates a broadband optical window and an active thermal management system, maintaining optical performance while effectively resisting high-pressure penetration and temperature gradient effects, thereby ensuring long-term operational stability of the optoelectronic system under complex hydrodynamic environments.

The main system parameters are listed in

Table 1. Main parameters of the UDHSL system.

Parameter	Description
Wavelength range	450–700 nm
Spectral bandwidth	10–300 nm (tunable)
Repetition rate	50 kHz–1 MHz (tunable)
Laser divergence angle	1 mrad
Pulse width *	7 ns
Laser pulse energy **	7.5 μJ
Echo sampling rate	10 GHz
Receiving optics diameter	202 mm
Scanning FOV	Azimuth: ±35°; Elevation: −5° to 20°
Working distance ***	≥27 m

* Full width at half maximum (FWHM); ** at 450–700 nm; *** at 450–550 nm with water attenuation coefficient of 0.05 m⁻¹.

.

2.2. Underwater Radiative Transfer Model

2.2.1. Hyperspectral LiDAR Underwater Radiative Transfer Equation

The backscattered echo signal received by a LiDAR system can be described by the general LiDAR equation [25,26], expressed as:

$$P_r={\displaystyle \frac{P_t{D_r}^2}{4\pi R^4{{\beta }_t}^2}}\mathsf{\sigma }{\eta }_{med}{\eta }_{sym}$$

(1)

where $P_r$ is the total received power at the sensor, $P_t$ is the transmitted laser power, $R$ is the distance between the target and the sensor, ${\beta }_t$ is the laser beam divergence angle (which is very small, ${\beta }_t\to 0$, $\sin {\beta }_t\approx {\beta }_t$), $D_r$ is the diameter of the receiving optical system, ${\eta }_{med}$ is the transmission factor of the medium, ${\eta }_{sym}$ represents the LiDAR system parameters, and $\mathsf{\sigma }$ is the backscattering cross-section of the target, which is related to the target reflectance properties. Let $\mathsf{\theta }$ (the laser incidence angle) denote the angle between the incident laser beam and the surface normal vector, $\mathsf{\Omega }$ the solid angle in the LiDAR backscattering direction, $A_s={\displaystyle \frac{\pi R^2\sin {{\beta }_t}^2}{4}}\approx {\displaystyle \frac{\pi R^2{{\beta }_t}^2}{4}}$ is the effective receiving area of the target, and $\rho$ the target reflectance.

For an ideal Lambertian diffuse reflector, the backscattering cross-section can be expressed as $\mathsf{\sigma }=\pi R^2{{\beta }_t}^2\rho cos\theta$ [25]. Substituting into Equation (1) and introducing wavelength dependence for the HSL system, the received intensity at wavelength $\lambda$ can be expressed as:

$$I_r(\lambda )={\displaystyle \frac{P_t(\lambda ){D_r}^2\rho (\lambda )cos\theta }{4R^2}}{\eta }_{med}(\lambda ){{\eta }_{sym}(\lambda )}^{\prime }$$

(2)

2.2.2. Influence of the Medium on Laser Transmission ${\eta }_{med}(\lambda )$

In the detection link of the UDHSL designed for underwater exploration, the laser beam traverses the media illustrated in Figure 2.

The laser transmission link passes through three media: air, glass, and water. The optical path lengths through air and glass are short, and their effects on the laser can be neglected. This study primarily discusses the influence of the water body on laser transmission.

The optical properties of water can be classified into inherent optical properties (IOPs) and apparent optical properties (AOPs) [27]. IOPs depend solely on the physical and optical properties of seawater itself, while AOPs are jointly determined by the IOPs of seawater and the distribution of ambient light radiation field within the water body. In underwater environments where the UDHSL operates, ambient light is negligible; thus, only the IOPs of seawater are considered herein. Laser light transmitted through seawater is subject to both absorption and scattering by the water body.

Laser attenuation in seawater is severe. Assuming a homogeneous water body along the beam transmission path, the change in laser intensity after propagating a distance R through seawater is typically described by the Beer–Lambert law [28,29]:

$${I_r\left(R,\lambda \right)=I}_t\left(\lambda \right)e^{-{\alpha }_w\left(\lambda \right)R}$$

(3)

where ${\alpha }_w\left(\lambda \right)$ is the attenuation coefficient (in m⁻¹), which is a function of wavelength, meaning that different wavelengths of light undergo different degrees of attenuation underwater. In practice, the water body along the beam path is not perfectly homogeneous; therefore, a correction factor $\mathsf{\xi }\left(\lambda ,R\right)$ is introduced. Considering the round-trip transmission through water:

$${\eta }_{med}\left(\lambda \right)=e^{-2{\alpha }_w\left(\lambda \right)R}·\mathsf{\xi }\left(\lambda ,R\right)$$

(4)

Furthermore, the attenuation coefficient can be expressed as the sum of the seawater absorption coefficient and the scattering coefficient ${\alpha }_w\left(\lambda \right)=a\left(\lambda \right)+b\left(\lambda \right)$, where $a\left(\lambda \right)$ is the wavelength-dependent absorption coefficient and $b\left(\lambda \right)$ is the wavelength-dependent scattering coefficient of seawater.

2.2.3. Underwater Radiative Transfer Characteristics

Incorporating the influence of the water medium, the HSL underwater radiative transfer equation (Equation (2)) can be written as:

$$I_r(\lambda )={\displaystyle \frac{P_t(\lambda )D_r^2\rho (\lambda )cos\theta }{4R^2}}{e}^{-2{\alpha }_w\left(\lambda \right)R}{\mathsf{\xi }\left(\lambda ,R\right)\eta }_{sym}(\lambda {)}^{\prime }$$

(5)

Assuming the object is an ideal Lambertian surface with reflectance independent of the incidence angle $cos\theta =1$, and that for a given wavelength the HSL system parameters $P_t(\lambda )$, $D_r$, and ${{\eta }_{sym}(\lambda )}^{\prime }$ are constants, the above equation simplifies to:

$$I_r(\lambda )=\mathrm{C}\left(\lambda \right)\rho (\lambda ){\displaystyle \frac{e^{-2{\alpha }_w\left(\lambda \right)R}}{R^2}}\mathsf{\xi }\left(\lambda ,R\right)$$

(6)

$$\mathrm{C}\left(\lambda \right)={\displaystyle \frac{D_r^2}{4}}{P}_t(\lambda ){{\eta }_{sym}(\lambda )}^{\prime }$$

(7)

where $\mathrm{C}\left(\lambda \right)$ depends on the system and the selected wavelength. For a single filtered wavelength band in a single laser pulse:

$$I_r\propto {\displaystyle \frac{e^{-2{\alpha }_{w}R}}{R^2}}·\rho ·\mathsf{\xi }\left(R\right)$$

(8)

The echo intensity depends solely on the distance and the target reflectance. For a fixed target at a fixed position:

$$I_r(\lambda )\propto {P_t(\lambda )e}^{-2{\alpha }_w\left(\lambda \right)R}{{\eta }_{sym}(\lambda )}^{\prime }\mathsf{\xi }\left(\lambda \right)$$

(9)

The echo intensity at different wavelengths is influenced by the laser source emission power at that wavelength, the water body attenuation coefficient at that wavelength, and the system transmittance at that wavelength.

In summary, the UDHSL underwater radiative transfer model (Equation (5)) is derived under the following assumptions: (i) the target is an ideal Lambertian diffuse reflector whose reflectance is independent of the incidence angle; (ii) the water body along the laser propagation path is optically homogeneous, such that the attenuation coefficient ${\alpha }_w\left(\lambda \right)$ is spatially constant over the detection distance; (iii) the optical paths through air and the pressure-resistant window glass are negligibly short compared to the underwater path, and their attenuation contributions are absorbed into the system parameter $\mathrm{C}\left(\lambda \right)$; (iv) the laser incidence angle $\theta$ is approximately zero (near-normal incidence); and (v) only single-scattering interactions are considered. The correction factor $\mathsf{\xi }\left(\lambda ,R\right)$ accounts for residual deviations from these idealized conditions. Equation (5) constitutes the complete radiative transfer model for the UDHSL system. Under these assumptions, Equations (6)–(9) provide simplified proportionality relationships that isolate the dependence of echo intensity on specific physical variables (reflectance, distance, and wavelength), forming the basis for the controlled single-variable experiments in Section 2.3. These assumptions are appropriate for the controlled laboratory environment. Under the homogeneous freshwater conditions of the present experiments, ξ(λ,R) ≈ 1 and is not explicitly estimated; it remains in the formulation to preserve generality for future applications in non-homogeneous waters environments of this study (static, homogeneous freshwater). Extension to vertically stratified, dynamically mixing, or highly turbid natural waters would require a depth-dependent attenuation profile and explicit treatment of multiple scattering, which are beyond the scope of the present work. Summary of key parameters in the UDHSL radiative transfer model is shown in

Table 2. Summary of key parameters in the UDHSL radiative transfer model.

Symbol	Parameter	Physical Meaning	Unit	Note
$P_t(\lambda )$	Transmitted power	Laser emission power at λ	W	$\lambda$-dependent
$D_r$	Aperture diameter	Receiving optics diameter	m	Constant
$\rho (\lambda )$	Reflectance	Target surface reflectance	—	0–1
$R$	Distance	Sensor-to-target distance	m	—
$\theta$	Incidence angle	Laser-to-surface-normal angle	rad	≈0 herein
${\alpha }_w\left(\lambda \right)$	Attenuation coeff.	$a\left(\lambda \right)+b\left(\lambda \right)$, total water attenuation	m−1	$\lambda$-dependent
$a\left(\lambda \right)$	Absorption coeff.	Water absorption coefficient	m−1	—
$b\left(\lambda \right)$	Scattering coeff.	Water scattering coefficient	m−1	—
$\mathsf{\xi }\left(\lambda ,R\right)$	Correction factor	Non-homogeneity correction	—	≈1 if homogeneous
$\mathrm{C}\left(\lambda \right)$	System parameter	Composite (Equation (7))	—	$D_r$, $P_t$, ${\eta }_{sym}$
${{\eta }_{sym}(\lambda )}^{\prime }$	System efficiency	Optical + detector efficiency	—	Incl. detector area
${\beta }_t$	Beam divergence	Half-angle divergence	rad	Small value

.

2.3. Underwater Detection Experiments

To verify the influence mechanisms of each contributing factor on the laser echo intensity as described by Equation (5), two sets of single-variable experiments were designed to obtain results under varying target reflectance and detection distance conditions, respectively.

Both experiments employed a horizontal laser propagation configuration rather than the downward-looking geometry typical of operational seabed mapping. This choice was made for the following reasons: (a) the horizontal configuration allows precise control of the target distance using calibrated mounting positions along the water pipe or pool gantry rail, which is essential for controlled single-variable experiments; (b) it avoids the confounding effects of depth-dependent hydrostatic pressure, temperature gradients, and vertical stratification of water optical properties, thereby isolating the specific variables under study (reflectance, distance, and wavelength); and (c) the Beer–Lambert-based radiative transfer model (Equation (5)) describes attenuation as a function of optical path length through the medium rather than of propagation direction, so horizontal laboratory results provide valid first-order validation of the model physics. Future work should include validation in vertical downward-looking configurations to address depth-stratification effects relevant to field deployments.

2.3.1. Reflectance Experiment

Standard diffuse reflectance panels with nominal reflectance values ranging from 20% to 90% were selected as experimental targets (Figure 3). The panels are Lambertian diffuse reflectors fabricated from spray-coated aluminum substrates with a specialized waterproof coating treatment, ensuring that their reflectance performance remains stable during short-term water immersion. The spectral reflectance of each panel in the 400–700 nm band is relatively flat, with spectral variation amplitude not exceeding 5%, meeting the requirements for broadband spectral measurements.

The reflectance of the panels was first measured in air using an SVC spectrometer (Spectra Vista Corporation, Poughkeepsie, NY, USA) to obtain reference reflectance data. Prior to underwater measurements, each panel surface was thoroughly pre-wetted with running water to displace surface air bubbles, which were observed to affect the apparent reflectance in preliminary tests. This pre-wetting procedure ensured that the panels were virtually free of surface bubbles upon submersion. After immersion, measurements commenced within approximately 30 min, minimizing the duration of water exposure. Subsequently, the panels were placed in a 27 m long opaque grey PVC pipe (inner diameter 315 mm) filled with fresh tap water. The pipe was equipped with an optical window on one end and multiple capped tee-junctions for target placement. The UDHSL system was positioned outside the optical window to acquire backscattering echo intensity from the standard panels in water (Figure 4 and Figure 5).

The experimental conditions were as follows: UDHSL wavelength range of 450–560 nm, spectral bandwidth of 10 nm, and laser repetition rate of 50 kHz. During the experiment, 20 backscattering echo measurements were recorded for each target at each wavelength, and the average value was taken as the intensity result to improve the signal-to-noise ratio. The detection distance was 4.01 m, and the water attenuation coefficient was approximately 0.05 m⁻¹ at 550 nm. The detection distance was measured using a Leica(Leica Geosystems, Heerbrugg, Switzerland) laser rangefinder with an accuracy of ±0.001 m. The water attenuation coefficient was measured using a WET Labs(Philomath, OR, USA) ACS absorption–attenuation meter.

2.3.2. Distance Effect Experiment

To thoroughly investigate the influence mechanism of distance on UDHSL echo signals, targets at varying distances were measured in two different water quality environments.

(1) Clear-water pipe distance experiment

First, measurements were conducted in the pipe using fresh tap water under the same conditions as described in Section 2.3.1. Standard diffuse reflectance panels were placed at detection distances of 4.01, 6.30, and 8.65 m in the pipe, and the UDHSL system was used for target detection.

(2) Turbid-water pool distance experiment

Subsequently, measurements were conducted in a large pool facility (50 m long, 15 m wide, 10 m deep). The pool walls and floor were fitted with black light-absorbing baffles to suppress stray light and multiple reflections. As shown in Figure 6, the pool was equipped with two independent gantry crane structures above the water surface, each with precision numerically controlled traversing systems mounted on their crossbeams. Computer-controlled displacement along guide rails enabled separate positioning of the measurement instrument and the target assembly. Each gantry was fitted with motorized telescopic poles to lower test components to a specified depth underwater.

The UDHSL system was encapsulated in a waterproof shell and rigidly connected to the east gantry telescopic pole via a flange, then vertically immersed to a depth of 2 m. An 80% reflectance standard panel was mounted on the west gantry at the same depth, maintaining coaxial alignment with the UDHSL system. The distance between the UDHSL and the target was varied using the numerically controlled traversing system.

The experimental conditions were as follows: UDHSL wavelength range of 510–590 nm, spectral bandwidth of 20 nm, laser repetition rate of 50 kHz, and detection distances of 4, 5, 6, 7, and 8 m. The water attenuation coefficient was approximately 0.3 m⁻¹ at 550 nm. Figure 7 shows the underwater test setup.

The pool contained natural water whose turbidity originated primarily from non-pigmented suspended particulate matter. The water optical properties were independently characterized using a WET Labs AC-S absorption–attenuation meter (spectral range 400–730 nm, spectral resolution ≈4 nm, optical path length 0.25 m), calibrated with ultrapure water prior to each measurement session. Spectral decomposition of the AC-S measurements using a parametric inherent optical property (IOP) model revealed that the non-water attenuation was dominated by particulate absorption (a_d(440) ≈ 0.40 m⁻¹, with exponential spectral decay), while particulate scattering was comparatively low (b_p ≈ 0.07 m⁻¹, spectrally near-flat), indicating an absorption-dominated water type. To assess spatial homogeneity, AC-S measurements were performed at multiple locations within the pool; the spatial variation in the attenuation coefficient was within ±10%. Measurements were also taken before and after each experimental session to monitor temporal stability.

Additionally, a custom-built hyperspectral LiDAR calibration device was used as a measurement target for underwater tests. The device contained four 65% reflectance diffuse panels arranged sequentially from front to back, connected by crossbeams. The device was rigidly attached to the UDHSL system via crossbeams. A pull rod mechanism enabled deployment and retraction of the panels, allowing the UDHSL to image panels at different positions across four configurations. The experimental conditions were identical to those described above. Figure 8 shows the schematic of the calibration device.

3. Results

3.1. Normalized Echo Intensity Data Processing Method

Since the optical window glass partially reflects the emitted laser, the raw full-waveform data acquired by the high-speed digitizer consists of two peaks, as illustrated in Figure 9 (shown at 550 nm). The first echo originates from the window glass reflection, while the second echo is the target backscattering return signal. Defining the window glass echo as the main wave in LiDAR offers two advantages: (1) the window glass constitutes the water entry interface, and the temporal position of the window glass echo serves as the ranging origin, enabling accurate calculation of the laser propagation distance in water; (2) since the reflectance of the window glass is constant, the window glass echo intensity is proportional to the total intensity of the current laser pulse, and normalizing the target echo eliminates errors caused by pulse-to-pulse energy fluctuations of the laser source:

$$I_{Normalizde}={\displaystyle \frac{I_{Echo}}{I_{Mainwave}}}$$

(10)

where $I_{Normalizde}$ represents the normalized echo intensity, $I_{Echo}$ denotes the target echo intensity, and $I_{Mainwave}$ is the window glass echo intensity (main wave intensity).

3.2. Relationship Between Echo Intensity and Target Reflectance

3.2.1. Reflectance Measurements of Standard Panels in Air

The reflectance of the standard diffuse reflectance panels (20–90%) was measured in air using an SVC spectrometer. The results are shown in Figure 10.

As shown in the figure, the actual reflectance of the standard panels exhibits some deviation from the nominal values.

Table 3. Statistical results of reflectance of standard diffuse reflectance panels measured in air.

Nominal Reflectance	Average Reflectance	Reflectance Deviation	RMSE
20%	22.48%	2.48%	2.49%
30%	32.86%	2.86%	2.89%
40%	45.33%	5.33%	5.35%
50%	60.12%	10.12%	10.24%
55%	70.54%	15.54%	15.57%
60%	74.31%	14.31%	14.32%
65%	76.49%	11.49%	11.52%
70%	79.21%	9.21%	9.29%
80%	85.47%	5.47%	6.70%
90%	99.88%	9.88%	10.35%

presents the average reflectance, reflectance deviation, and root mean square error (RMSE) of each panel in the 400–700 nm wavelength range. These measured reflectance values are used in Section 3.2.2 as an independent reference for a parallel intensity–reflectance regression.

3.2.2. Reflectance Experiment Results in Clear Water Pipe

Using the experimental setup shown in Figure 4, standard diffuse reflectance panels with nominal reflectance values of 20–90% were placed at a distance of 4.01 m in the clear-water pipe, and multi-band scanning was performed using the UDHSL system. Twenty echo waveforms were collected per wavelength band, covering the wavelength range of 450–680 nm with a 10 nm interval.

For the acquired echo waveform data, peak detection algorithms were first applied to identify the peak positions and intensities of the main wave (window glass echo) and the target echo. The detection distance was calculated from the time difference between the main wave and the target echo, and anomalous data with distance deviation exceeding ±0.2 m were excluded. Additionally, waveforms with echo peak values exceeding 95% of the 16-bit digitizer full-scale range (i.e., unsigned peak > 62,000 out of 65,535 counts) were identified as saturated and excluded. Noise estimation and denoising processing were performed on the waveform data, and the normalized echo intensity was calculated using Equation (10). Figure 11 shows the normalized echo intensity as a function of wavelength for each reflectance panel.

As shown in the figure, the normalized echo intensity generally increases with increasing nominal reflectance and is relatively high in the 450–550 nm wavelength range, gradually attenuating with increasing wavelength. This spectral trend is consistent with the wavelength-dependent attenuation predicted by Equation (9), experimentally validating the radiative transfer model under the specific instrument and water conditions of this study.

Table 4. Normalized echo intensity of each reflectance panel in the water pipe (450–680 nm, 10 nm step, detection distance 4.01 m).

Nominal Reflectance	Mean Normalized Echo Intensity	Standard Deviation	Spectral Var. (%)	Repeat Var. (%)
20%	2.8685	1.5350	94.4	5.6
30%	5.2531	3.2176	92.7	7.3
40%	4.8450	2.7317	94.4	5.6
50%	6.7942	4.0432	90.1	9.9
55%	8.0931	4.4885	87.4	12.6
60%	8.8501	5.3735	90.6	9.4
65%	10.2223	6.9988	77.1	22.9
70%	13.3811	9.4522	80.5	19.5
80%	13.4802	15.1204	59.1	40.9
90%	16.9388	13.4511	45.6	54.4

summarizes the average normalized echo intensity and standard deviation for each reflectance panel across all wavelength bands. The reported standard deviation combines inter-band spectral variability and intra-band measurement repeatability; a variance decomposition of these two components is given in the last two columns of Table 4.

The linear fitting result demonstrates a strong positive linear correlation between the average normalized echo intensity and the nominal target reflectance, with a fitting equation of y = 0.198x − 2.005 and a coefficient of determination R² = 0.938. This indicates that at a 4 m underwater detection distance, the normalized echo intensity of the UDHSL system effectively reflects differences in target reflectance. The error bars in the figure represent the standard deviation of measurements across all wavelength bands, revealing that high-reflectance panels (80% and 90%) exhibit greater measurement fluctuations, as the stronger echo signals from high-reflectance targets are proportionally more sensitive to environmental perturbations. Figure 12a presents the regression using nominal reflectance as the independent variable (R² = 0.938); Figure 12b presents the parallel regression using SVC-measured actual reflectance (Table 3) as the independent variable (R² = 0.873). Both regressions confirm a strong positive linear relationship between normalized echo intensity and target reflectance. The lower R² in (b) is attributed to three factors: (i) the inherent measurement uncertainty of the SVC field spectrometer under ambient conditions; (ii) the Fresnel interface effect at the coating–medium boundary, which modifies the effective panel reflectance upon submersion; and (iii) the compressed spacing of SVC-measured values for the mid-to-high reflectance panels. A normal-incidence Fresnel estimate indicates that immersing a diffuse panel (coating refractive index n ≈ 1.5–2.0) from air (n = 1.0) into water (n = 1.33) reduces the interface reflection loss and increases the effective reflectance by approximately 5–17%, contributing to the discrepancy between air-measured and underwater effective reflectance.

Table 5. Normalized echo intensity of each reflectance panel indexed by SVC-measured reflectance (450–680 nm, 10 nm step, detection distance 4.01 m).

SVC-Measured Reflectance (%)	Mean Normalized Echo Intensity	Standard Deviation	Spectral Var. (%)	Repeat Var. (%)
22.48%	2.8685	1.5350	94.4	5.6
32.86%	5.2531	3.2176	92.7	7.3
45.33%	4.8450	2.7317	94.4	5.6
60.12%	6.7942	4.0432	90.1	9.9
70.54%	8.0931	4.4885	87.4	12.6
74.31%	8.8501	5.3735	90.6	9.4
76.49%	10.2223	6.9988	77.1	22.9
79.21%	13.3811	9.4522	80.5	19.5
85.47%	13.4802	15.1204	59.1	40.9
99.88%	16.9388	13.4511	45.6	54.4

presents the same statistics indexed by the SVC-measured actual reflectance (Table 3), providing a direct look-up for the error bars in Figure 12b.

Based on the proportionality predicted by Equation (8), linear regressions were performed between the normalized echo intensity and target reflectance, using both nominal and SVC-measured values as the independent variable, and on a per-wavelength basis. Figure 12a,b show the wavelength-averaged results using nominal (R² = 0.938) and measured (R² = 0.873) reflectance, respectively; Figure 12c presents R² values for each of the 22 individual bands, of which 16 exceed 0.85 (mean R² = 0.915 in the 500–660 nm core range). Figure 12d shows the corresponding analysis using SVC-measured reflectance, yielding a mean per-band R² of 0.789 with 9 of 22 bands exceeding 0.85. The lower R² below 490 nm reflects the weak supercontinuum source power in that spectral region rather than a departure from the reflectance proportionality of Equation (8). The error bars in (a) and (b) correspond to the standard deviations listed in Table 4 and Table 5, with the 80% and 90% panels exhibiting larger standard deviations, as the stronger echo signals from high-reflectance targets are proportionally more sensitive to environmental perturbations such as optical-path fluctuations and ambient scattering variations.

3.3. Relationship Between Echo Intensity and Detection Distance

3.3.1. Clear-Water Pipe Distance Experiment Results

To further investigate the effect of detection distance on UDHSL normalized echo intensity, three detection distances (4.01, 6.30, and 8.65 m) were set in the water pipe, and echo detection was performed on 10 standard diffuse reflectance panels (nominal reflectance: 20–90%). The data processing method was identical to that described in Section 3.2.2. Normalized echo intensity was calculated for each wavelength band using Equation (10), and the average across all valid bands was computed to obtain the mean normalized echo intensity for each reflectance panel at each distance. Figure 13 shows the variation in normalized echo intensity with detection distance for each reflectance panel.

As shown in Figure 13 and

Table 6. Mean normalized echo intensity of each reflectance panel at different detection distances (clear-water pipe, 450–680 nm, 10 nm step).

Nominal Reflectance	4.01 m	6.30 m	8.65 m
20%	2.8685	0.7522	0.3784
30%	5.2531	2.1893	0.9416
40%	4.8450	1.5992	0.6917
50%	6.7942	2.4322	1.0438
55%	8.0931	2.7472	1.2244
60%	8.8501	3.5484	1.4174
65%	10.2223	3.8217	1.4329
70%	13.4811	5.6763	1.9805
80%	13.4802	5.6180	2.0824
90%	16.9388	5.7577	2.2317

, for all reflectance panels, the normalized echo intensity generally decreases with increasing detection distance, consistent with the theoretical prediction of the LiDAR equation in which received power is inversely proportional to the square of the distance. Taking the 90% nominal reflectance panel as an example, its normalized echo intensity decreases from 16.94 at 4.01 m to 2.23 at 8.65 m, representing an attenuation of over 86%. The difference in normalized echo intensity among panels of different reflectance values diminishes with increasing distance: the intensity difference between the highest (90%) and lowest (20%) reflectance panels is approximately 14.07 at 4.01 m, but narrows to 1.85 at 8.65 m. This indicates a reduced capability of the system to discriminate between targets of different reflectance at longer detection distances.

3.3.2. Turbid-Water Pool Distance Experiment Results

To investigate the distance detection capability of the UDHSL system under turbid-water conditions, echo measurement experiments were conducted at different detection distances in a large pool. Under the turbid water conditions, the standard diffuse reflectance panel was placed at distances of approximately 4, 5, 6, 7, 8, and 10 m from the UDHSL system. The laser wavelength covered 450–650 nm (step size: 20 nm), and 10 echo signals were collected per wavelength band. Data processing followed the same procedure as described in Section 3.2.2. Figure 14 shows the normalized echo intensity as a function of wavelength for each detection distance.

As shown in Figure 14 and

Table 7. Mean normalized echo intensity at different wavelengths and detection distances in the turbid-water pool (80% nominal reflectance panel).

Wavelength	4 m	5 m	6 m	7 m	8 m	10 m
450 nm	–	0.1785	0.0459 *	0.1655 *	–	–
470 nm	1.1324	0.1314	0.0451	–	–	–
490 nm	1.3739	0.2363	0.0547	0.0168	–	–
510 nm	1.6237	0.3161	0.0860	0.0169	–	0.0032 *
530 nm	1.8166	0.3976	0.1207	0.0278	0.0082 *	–
550 nm	1.8017	0.4480	0.1417	0.0389	0.0083	0.0056 *
570 nm	1.6379	0.4714	0.1531	0.0445	0.0126	–
590 nm	1.4505	0.3286	0.0944	0.0204	–	–
610 nm	0.8918	0.1442	0.0236	0.0098 *	–	–
630 nm	0.8700	0.1195	0.0215	–	–	–
650 nm	0.7945	0.0783	–	0.0238 *	–	–

* Values marked with an asterisk indicate fewer than 3 valid data points, with lower statistical reliability. “–” indicates no valid echo signal detected.

, the attenuation effect of turbid water on UDHSL echo intensity is highly pronounced. At a detection distance of 4 m, the normalized echo intensity peaks in the 530–570 nm band (approximately 1.64–1.82), exhibiting a distinct water transmission window characteristic. As the detection distance increases to 5 m, the normalized echo intensity decreases dramatically to approximately 0.27, representing an attenuation of about 80%. At detection distances of 6 and 7 m, the normalized echo intensity further decreases to 0.09 and 0.03, respectively, with the echo signal approaching the system noise floor. At 8 and 10 m, only a few wavelength bands (530–570 nm) yield extremely weak echo signals, with normalized intensity values below 0.01 and 0.005, and very few valid data points. Notably, the 4 m data exhibit substantially larger error bars than other distances (coefficient of variation ≈35–40% in the 530–570 nm bands, compared with ≈11–13% at 5 m); the origin of this near-field variability is discussed in Section 4.

From the wavelength perspective, the spectral peak at 530–570 nm is consistent with the minimum water attenuation coefficient in this region. As the detection distance increases, the wavelength range over which valid echo signals can be detected progressively narrows: from 470–650 nm at 4 m, to 490–630 nm at 6 m, and only 510–590 nm at 7 m. This progressive spectral narrowing is consistent with the wavelength-dependent exponential attenuation predicted by Equation (5): bands farther from the minimum-attenuation window have larger c(λ) values and therefore experience faster signal decay, causing their echoes to fall below the detection threshold at shorter propagation distances.

Compared with the clear-water pipe experiment results in Section 3.3.1, the distance attenuation effect in the turbid water environment is significantly enhanced. Under clear water conditions, the UDHSL system still obtains normalized echo intensities of approximately 0.38–2.23 at 8.65 m (depending on reflectance); in contrast, in turbid water, the normalized echo intensity has already decreased to 0.09 at 6 m and further to 0.03 at 7 m. This demonstrates that water turbidity is the critical factor limiting UDHSL detection range, as scattering and absorption by suspended particles cause rapid echo signal attenuation. The effective detection range of the system in turbid water is approximately 5–6 m, substantially less than the >10 m achievable under clear water conditions.

3.4. Quantitative Validation of the Attenuation Model

To provide a direct quantitative test of the radiative transfer model, the normalized echo intensity data from the clear-water pipe (Section 3.3.1) and turbid-water pool (Section 3.3.2) experiments were fitted to the exponential decay model derived from Equation (5). For each reflectance panel (clear water) and wavelength band (turbid water), the beam attenuation coefficient c(λ) was extracted via nonlinear least-squares fitting of the form I = A·exp(−2c(λ)R)/R². The fitted values were then compared with independent measurements obtained using a WET Labs AC-S absorption–attenuation spectrometer.

Table 8. Comparison of fitted and independently measured (AC-S) beam attenuation coefficients at 550 nm.

Experiment	Fitted	AC-S	Δ	Relative Error
Clear water	0.031	0.028	+0.003	+11%
Turbid water	0.462	0.447	+0.015	+3.4%

summarizes the comparison at 550 nm.

As shown in Table 8, the fitted attenuation coefficients agree closely with the AC-S reference values, with deviations of +0.003 m⁻¹ (+11%) for clear water and +0.015 m⁻¹ (+3.4%) for turbid water. Figure 15 further confirms the validity of the exponential decay model: plots of ln(I·R²) versus R yield highly linear relationships (R² > 0.99 for all conditions), with slopes corresponding to −2c(λ), as predicted by the Beer–Lambert law.

To characterize the full spectral dependence of the water attenuation, Figure 16 compares the beam attenuation coefficients measured by the AC-S (WET Labs, path length 0.25 m) with those retrieved from the Beer–Lambert inversion of the UDHSL echo signals. For the clear water, the broadband inversion yields a mean attenuation coefficient of 0.039 m⁻¹, in agreement with the AC-S measurement at 550 nm (c(550) = 0.046 m⁻¹). For the turbid water, the per-wavelength inversion results across 490–590 nm closely track the AC-S spectral curve, with a deviation of less than 2% at 550 nm (α = 0.334 m⁻¹ versus c = 0.340 m⁻¹). The close spectral agreement confirms that the attenuation coefficients derived from the UDHSL measurements are consistent with independent AC-S observations across the visible spectrum.

4. Discussion

This study systematically validated the underwater hyperspectral LiDAR radiative transfer model (Equations (5)–(9)) through underwater measurements using standard diffuse reflectance panels. Equation (8) indicates that for a single laser pulse at a given wavelength, the normalized echo intensity depends solely on the detection distance and the target reflectance. The linear fitting result in Section 3.2.2 (y = 0.198x − 2.005, R² = 0.938; R² = 0.873 using SVC-measured reflectance) confirms a strong positive linear correlation between normalized echo intensity and nominal target reflectance under fixed-distance conditions, validating the theoretical derivation of Equation (9) that the echo intensity is constant for a fixed target at a fixed position. The R² value slightly below unity may be attributed to the following factors: (1) deviation between the actual and nominal reflectance values of the standard panels (Table 3), and (2) the stronger echo signals from high-reflectance panels (80% and 90%) being proportionally more susceptible to environmental perturbations such as optical-path fluctuations and ambient scattering variations.

The Beer–Lambert law in Equation (3) dictates that laser intensity decays exponentially as it propagates through water. The two distance effect experiments in Section 3.3 verify this relationship at the experimental level. In the clear-water pipe experiment (attenuation coefficient ≈0.05 m⁻¹ at 550 nm), the normalized echo intensity exhibits a pronounced decreasing trend with increasing distance. For the 90% reflectance panel, the intensity decreases from 16.94 at 4.01 m to 2.23 at 8.65 m, an attenuation exceeding 86%. In the turbid-water pool experiment (attenuation coefficient ≈0.3 m⁻¹ at 550 nm), the distance attenuation effect is dramatically enhanced: an approximately 80% intensity loss occurs with only a 1 m increase from 4 to 5 m, and at distances beyond 6 m the echo signal approaches the noise floor. The comparison between these two experiments demonstrates that the water attenuation coefficient is the critical parameter governing the effective detection range of the UDHSL, consistent with the theoretical prediction of the exponential decay term exp(−2c(λ)R) in Equation (5). Furthermore, the quantitative validation in Section 3.4 demonstrates that the fitted attenuation coefficients agree with independent AC-S measurements to within +11% (clear water) and +3.4% (turbid water), confirming the applicability of the Beer–Lambert exponential decay model.

Equation (9) indicates that echo intensity at different wavelengths is influenced by the combined effects of laser source emission power, water body attenuation coefficient, and system transmittance at each wavelength. The experimental results fully demonstrate this wavelength dependence. Under clear water conditions (Section 3.2.2), the normalized echo intensity is higher in the 450–550 nm band and gradually decreases with increasing wavelength, reflecting the inherent optical property that water has a smaller absorption coefficient in the blue–green spectral region [28,29]. Under turbid water conditions (Section 3.3.2), the spectral peak of the echo intensity shifts to 530–570 nm, and as detection distance increases, the detectable wavelength range narrows from 470–650 nm to 510–590 nm. This occurs because the scattering effect of suspended particles in turbid water is superimposed on the water absorption effect, further increasing the attenuation coefficient at marginal wavelengths and causing the signal to attenuate to undetectable levels within shorter transmission distances. These results suggest that the operational wavelength range of the UDHSL should be adjusted according to water quality conditions to maximize the signal-to-noise ratio and effective detection range.

The window glass echo normalization method adopted in this study (Equation (10)) offers two advantages: (1) the window glass echo serves as the ranging origin, enabling accurate calculation of laser propagation distance in water; and (2) normalization eliminates systematic errors caused by pulse-to-pulse laser energy fluctuations. However, this method has certain limitations. First, normalization cannot fully eliminate wavelength-dependent system differences (e.g., filter transmittance and detector responsivity), resulting in the spectral characteristics of the normalized echo intensity containing superimposed information from both water attenuation and system response. Second, under long-range or turbid water conditions, the target echo signal becomes weak or is buried in noise, reducing the reliability of normalization results. Future work may consider introducing system radiometric calibration parameters to further correct the normalized results at each wavelength, thereby more accurately retrieving the intrinsic target reflectance.

It should be noted that all experiments in this study were conducted in freshwater environments (tap water and turbid pool water). Seawater has fundamentally different inherent optical properties (IOPs) compared to freshwater, primarily due to dissolved salts, colored dissolved organic matter (CDOM), phytoplankton pigments, and suspended particulate matter. These constituents significantly modify the spectral absorption and scattering profiles, particularly in the blue–green region (380–550 nm) where CDOM absorption is strongest. While the Beer–Lambert law framework underlying our radiative transfer model (Equation (3)) is applicable to both freshwater and seawater, the specific values of the attenuation coefficient and the correction factor will differ substantially in marine environments. Therefore, the quantitative results reported herein (e.g., effective detection ranges and optimal wavelength bands) should be interpreted as freshwater-specific benchmarks and cannot be directly extrapolated to seawater without further validation. Future experiments in actual or simulated seawater environments are planned to extend the model validation to marine applications.

Regarding the potential influence of the pipe geometry on radiometric accuracy in the clear-water experiments (Section 2.3.1): the laser beam diameter at the maximum experimental distance of 8.65 m is approximately 14 mm (given the 1 mrad divergence and ~5 mm initial beam diameter), occupying less than 5% of the 315 mm pipe inner diameter. The beam therefore propagates well within the central region of the pipe without illuminating the walls. Furthermore, background measurements with no target present yielded no detectable echo signal above the noise floor, confirming that parasitic returns from wall scattering were negligible under clear-water conditions. The grey PVC inner surface also has inherently low reflectance, further limiting any wall-scattered contribution. For the turbid-water experiments, the open pool facility equipped with black light-absorbing baffles on all walls and the floor effectively suppressed stray reflections, providing a well-controlled optical environment.

In terms of engineering reference value and application boundaries, the present study should be positioned as: (a) a laboratory validation of the radiative transfer model for a new instrument class (UDHSL), demonstrating the fundamental correctness of the wavelength-dependent underwater LiDAR equation under controlled conditions; (b) a demonstration of the normalized echo intensity processing method that enables comparison of echo signals across wavelengths and distances; and (c) a quantitative characterization of the coupled distance–wavelength–reflectance effects under two controlled water quality conditions (clear freshwater and turbid freshwater). While these results do not directly prescribe operational parameters for field deployments, they establish the necessary theoretical and experimental foundation upon which field-specific calibration protocols can be built. The wavelength selection guidance (e.g., 450–550 nm for clear water, 530–570 nm for turbid water) should be interpreted as first-order engineering references that will require refinement when applied to specific marine environments with different inherent optical properties.

The elevated near-field intensity variability observed at 4 m in the turbid-water experiment (Section 3.3.2) also carries implications for future submersible-mounted UDHSL operations. In the present laboratory setup, the instrument’s water-cooling system continuously circulated ambient water through the waterproof shell, generating localized turbulence and temperature microstructure that significantly increased pulse-to-pulse intensity fluctuations in the near-field region (coefficient of variation ≈35–40% at 4 m vs. ≈11–13% at 5 m). In operational deployments on submersible platforms, analogous disturbances will arise from propulsion thrusters, thermal management exhausts, and ballast exchange flows. Potential mitigation strategies include extended mounting booms to distance the sensor from platform-induced turbulence, hydrodynamic shielding structures around the optical aperture, and measurement protocols that suspend data acquisition during active thruster operation.

5. Conclusions

Targeting the demand for synchronous acquisition of underwater terrain and material information, this study developed a UDHSL system with a wavelength range of 450–700 nm and a maximum detection distance exceeding 27 m. Building upon the conventional LiDAR equation, a wavelength-dependent water attenuation correction term was introduced to establish a hyperspectral LiDAR radiative transfer model applicable to optically homogeneous underwater environments under single-scattering conditions. Through systematic experiments using standard diffuse reflectance panels in air, clear water, and turbid water, the following principal conclusions were obtained:

(1) Reflectance effect: At a 4.01 m clear-water detection distance, the normalized echo intensity of 10 reflectance panels (nominal reflectance: 20–90%) exhibits a linear positive correlation with the nominal reflectance, with a fitting equation of y = 0.198x − 2.005 and R² = 0.938 (R² = 0.873 using SVC-measured reflectance). This demonstrates that the normalized echo intensity of the UDHSL system effectively characterizes reflectance differences among underwater targets, validating the linear relationship between echo intensity and target reflectance predicted by the radiative transfer model.

(2) Distance attenuation effect: In the clear-water pipe experiment (attenuation coefficient ≈0.05 m⁻¹ at 550 nm), the normalized echo intensity attenuates significantly with increasing distance, exceeding 86% from 4.01 to 8.65 m, while the intensity difference among panels of different reflectance values converges with distance, indicating reduced reflectance discrimination capability at longer ranges. In the turbid-water pool experiment (attenuation coefficient ≈0.3 m⁻¹ at 550 nm), the distance attenuation is dramatically enhanced, reaching 80% from 4 to 5 m, with an effective detection range of approximately 5–6 m—substantially less than the >10 m achievable in clear water. The comparison between the two experiments confirms that the water attenuation coefficient is the key parameter governing the UDHSL detection range.

(3) Wavelength dependence: The echo intensity exhibits significant wavelength selectivity, with the highest response at 450–550 nm in clear water and the peak shifting to 530–570 nm in turbid water. As detection distance increases, the effectively detectable wavelength range progressively narrows, with only the 530–570 nm band yielding reliable echo signals at 7 m in turbid water. These findings provide an experimental basis for selecting optimal working wavelength bands for the UDHSL system under different water quality conditions.

The experiments reported herein validate the fundamental correctness of the underwater radiative transfer model; however, several directions warrant further investigation. First, the current normalization method does not fully eliminate wavelength-dependent system response differences, and future work could introduce system radiometric calibration parameters for absolute correction of echo intensity at each wavelength to enable intrinsic target reflectance retrieval. Second, the present experiments employed standard diffuse reflectance panels as targets under relatively idealized conditions. Recent studies have demonstrated that multispectral and full-waveform LiDAR techniques can effectively classify underwater targets such as mineral ores [18], sediment types [19], and submerged objects [20] under laboratory conditions; extending similar investigations to natural seabed substrates (e.g., sand, mud, rock, and coral) with the UDHSL system is an important next step to evaluate its material identification capability under complex target conditions. Third, to address the challenge of large dynamic range in echo intensity—where weak signal detection and strong signal saturation coexist—adaptive power control and wide dynamic range acquisition schemes should be developed. Fourth, by incorporating real-time measurement of water attenuation coefficients, a joint distance–water quality correction model can be established to improve reflectance retrieval accuracy under varying water quality conditions, thereby advancing the application of underwater hyperspectral LiDAR technology in underwater topographic mapping and seabed material identification. Finally, the current model assumes optically homogeneous water and single-scattering conditions; extending the radiative transfer framework to vertically stratified or multiple-scattering environments would broaden its applicability to field deployments.