1 Introduction

The oceans play a crucial role in regulating climate, supporting marine ecosystems, and influencing weather patterns. Understanding the global environmental parameters of the sea surface and subsurface is essential for studying oceanic processes, coastal erosion, marine ecosystem health, and climate change impacts. Remote sensing techniques, such as spaceborne devices, airborne instruments, and shipborne in-situ sensors, have emerged as powerful tools for collecting high-resolution profile data over large areas, thereby providing valuable insights into various oceanic parameters (McClain 2009; Chen et al. 2019a, b; Liu et al. 2022). Traditional remote sensing of ocean color based on spaceborne passive visible sensors has provided a deep view of the horizontal distribution of ocean optical properties and biogeochemical parameters; however, the lack of vertical information on the upper ocean layer limits further ocean study (Werdell et al. 2018). Spaceborne active remote sensing has been an efficient technique to illustrate the global vertical distributions of diel vertical migration (DVM) animals, with the most significant finding being that the animals arrive in the surface ocean at night. These results provide a valuable view of global DVM activities and develop a novel method for quantifying their biogeochemical abundance (Behrenfeld et al. 2019, 2022).

Airborne oceanic lidar has revolutionized our ability to study the fine-scale detail marine environmental parameters. Its applications in coastal zone mapping and monitoring of oceanic ecosystems have provided valuable insights into the dynamics of our oceans (Roddewig et al. 2017; Janowski et al. 2022). With ongoing advancements in lidar technology and data analysis techniques, airborne oceanic lidar is poised to play an increasingly significant role in understanding and managing our precious marine environment (Chen et al. 2021a; Song et al. 2021). Significant progress has been made in developing airborne oceanic lidar systems (Li et al. 2020a, b). Advanced laser technology, receiver design, and data processing algorithms have improved the efficiency and accuracy of data acquisition and analysis (Guo et al. 2021; Zhong et al. 2021). These advancements have opened up new possibilities for studying the marine environment with unprecedented detail and precision.

Airborne oceanic lidar is being extensively used for coastal ecosystem monitoring. By examining the distribution and behavior of marine organisms, such as phytoplankton, zooplankton, and fish, researchers can assess the health of marine ecosystems and understand their response to environmental changes. This information is crucial for sustainable fisheries management, biodiversity conservation, and preserving marine habitats (Churnside et al. 1997; Andrew et al. 2014).

The future of airborne oceanic lidar research holds great promise. Integrating lidar data with other remote sensing techniques, such as satellite imagery, will enhance the accuracy and coverage of oceanic studies (Popescu et al. 2011; Montgomery et al. 2019). The improved data processing and analysis algorithms will allow for better interpretation of oceanic parameters (Liu et al. 2020a, b). Additionally, deploying airborne oceanic lidar on unmanned aerial vehicles (UAVs) will provide more flexibility and cost-effectiveness in data acquisition, particularly in remote or inaccessible areas.

In this report, we reviewed the previous studies related to the remote sensing of the oceanic environmental parameters based on airborne lidar. Section 2 describes the configuration of the typical airborne oceanic lidar system. Section 3 illustrates the lidar data preprocessing and inversion model, including a step-by-step procedure for data post-processing. Section 4 presents the applications of the airborne oceanic lidar, including the oceanic optical and oceanic environmental parameters. Section 5 discusses the way ahead of the airborne oceanic lidar technique.

2 Typical airborne lidar configuration

2.1 Classification of airborne oceanic lidar

According to the basic principles and applications, airborne oceanic lidar can be categorized into airborne lidar altimeter or bathymetry (Yan et al. 2014), airborne elastic scattering lidar (Tuell et al. 2005; Churnside and Donaghay 2009), and airborne nonelastic scattering lidar (Liu et al. 2018a, b; Chen et al. 2021b) as listed in Table 1. Laser beams of the airborne lidar system penetrate through the ocean surface, water body, and bottom; therefore, the different parameters can be obtained from the backscattering echo signals.

Table 1 Airborne lidar classification and the measurement of remote sensing
Full size table

2.2 Configuration of typical airborne oceanic lidar

A typical airborne oceanic lidar system consists of a laser transmitter, multi-channel photon detectors, and a digitizer, as shown in Fig. 1. According to the function of different oceanic lidar, the transmission laser wavelength is either blue, green, or both. The receivers are also divided into channels with different wavelengths, like parallel polarization and cross-polarization.

Fig. 1
figure 1

Schematic diagram of a typical polarized oceanic lidar system

Full size image

The laser transmitter is the core device of airborne lidar. Current airborne oceanic lidar systems are normally equipped with mature and robust 532 nm lasers, providing as high as multi-kilohertz repetition rate and megawatt peak power laser pulses for coastal water detection. The most typical way is frequency doubling of a diode-pumped 1064 nm all-solid-state pulse laser (Tang et al. 2009; Lu et al. 2013, 2018). The performance of oceanic lidar systems with 470–490 nm is 1.5–1.75 times better than that with 532 nm in clean ocean waters (Gray et al. 2015).

Fourfold frequency conversion of 1.9 μm fiber laser is a practicable way to obtain 470–490 nm blue laser. Lockheed Martin and British Aerospace Systems have conducted relevant research (Honea et al. 2013; Creeden et al. 2016). Furthermore, frequency doubling of 0.9 μm quasi-three-level laser output of neodymium-doped yttrium aluminum garnet laser can achieve millijoule-level blue laser (Lu et al. 2019). Since 2018, a 486 nm optical parametric oscillator (OPO) based on beta barium borate (BBO) crystals and pumped by a 355 nm pulse laser has been developed, with a maximum pulse energy of 162 mJ and a peak power of 16.9 MW (Ma et al. 2020; Zhang et al. 2021a, b).

The return signal usually includes signals of the ocean surface, shallow-water channel, and deep-water to reach the demand of the large dynamic range of lidar echo from the sea surface to the subface or bottom. According to different applications, the signal channels may be divided into cross-polarization and co-polarization, fluorescence, Raman channels, etc. High quantum efficiency in blue-green wavelength is selected in airborne oceanic lidar, photomultiplier tube (PMT), and avalanche photodiode detector (APD) with fast response. The signal from the detector should be digitized with a high sampling rate and resolution to get high resolution.

The locations of the sea surface are usually obtained by identifying the maximum peak position of the echo signal data after the geometric and attitude correction (Liu et al. 2009). The profiles that contain the scattering information could be retrieved by the vertical range bins below the sea surface. Figure 2 depicts a typical example of data processing (Chen 2007). The lidar impulse response was reported in the previous study (Shen et al. 2022). The correct attenuated backscatter profile should be proceeded by removing the effects of the lidar impulse response (Lu et al. 2014; Lu et al. 2020). Generally, to prevent the direct reflection of the ocean surface, the observation angle of the lidar is usually inclined several degrees to the nadir.

Fig. 2
figure 2

Typical example of data processing. a Raw lidar data; b data after preprocessing. Figure redrawn and modified from Chen et al. (2021a), used with permission

Full size image

3 Algorithms

3.1 Radiative transfer simulation for airborne oceanic lidar

The lidar radiative transfer (RT) simulation aims to understand the emission, absorption, and scattering of laser in the medium environment. It attempts to model and ultimately quantify the intensity distribution of laser radiation in the medium. relationship between the physical parameters of media such as seawater, sea surface, seabed, and atmosphere and the process of laser energy transfer can be better understood by examining the interaction between laser and medium. This is a necessary technical support for oceanic lidar remote sensing research. One of the current priorities in oceanic lidar remote sensing is to improve the modeling of lidar multiple scattering effects under the influence of instrument and environmental parameters and to enhance the performance of the lidar RT model. Designing oceanic systems and inversion algorithms requires an effective simulation of the lidar RT process in seawater, an analysis of the accuracy of lidar system measurements under different environmental conditions, and an interpretation of the effects of observation geometry.

The existing methods for lidar RT simulation primarily include the lidar equation method (Gordon 1989; Abdallah et al. 2013; Comerón et al. 2017; Zhang et al. 2022), the standard Monte Carlo (MC) method (Gordon 1982; Ramella-Roman et al. 2005; Liang et al. 2006; Gabriel et al. 2013; Szczap et al. 2013; Dong et al. 2017; Szczap et al. 2021), the semi-analytical MC (SAA) method (Chen et al. 2019b; Liu et al. 2019; Zhu et al. 2019; Liu et al. 2020b; Chen et al. 2021a; Zhou et al. 2021; Chen et al. 2023), and the small-angle approximation (SAA) analytical method (Walker and McLean 1999; Hogan 2008; Kopilevich et al. 2010; Zhang et al 2021a, b).

Each of the methods mentioned above presents advantages and limitations. The lidar equation method solely considers single scattering and does not consider the occurrence of lidar multiple scattering in seawater. The standard MC method is relatively easy to implement but demands substantial computational resources, which is time and labor-consuming and introduces random noise in the computation. Conversely, the SAA method provides fast and robust performance, but it is primarily suited to specific application scenarios, ignoring large-angle scattering of particles and exhibiting limited adaptability to water bodies with high concentrations of subsurface phytoplankton. In contrast to the standard MC approach, the semi-analytic MC method incorporates a variance reduction technique that combines analytic calculations with stochastic MC calculations. This integration significantly improves computational efficiency and reduces noise (Fig. 3). Specifically, this method involves analytically calculating the probability at each collision that the photon will directly return to the receiver without additional interactions. Thus, this approach effectively reduces variance using computational and analytical techniques. Specifically, the method simulates a lidar system by incorporating geometric constraints to analytically estimate the probability of a remote receiver collecting a photon at a specific point (Chen et al. 2021a, b).

Fig. 3
figure 3

Comparisons between the standard MC, lidar equation (LQ), and semi-analytical MC under very large and small receiver FOVs

Full size image

The overall semi-analytic MC method flow is as follows: in the first step, one of the numerous photons was launched into the sea water. The water optical properties profile is divided into multiple layers, corresponding to the typical vertical resolution of the lidar system. Afterward, the photon proceeds a certain distance, and its position and weight are updated accordingly. Once the step is completed, the position of the photon is examined to determine whether it remains in the water within the FOV and detection area of the receivers or has escaped to the air or reflected off the sea bottom. If the photon is still within the water, it is scattered by a particle, and we calculate its current weight while determining if it is still alive. When the weight of the photon drops below a threshold value, usually around 10−4, the photon is considered dead. However, if the photon is still alive, it continues to move a distance, and we repeat the steps above until it meets its demise. Finally, the simulation results for all photons are aggregated. Compared to the standard MC model, the semi-analytic model provides an analytical estimation of the probability of photon collision, considering the geometric constraints imposed by the lidar system in both homogeneous and stratified waters.

3.2 Retrieval method of airborne oceanic lidar

The contribution of marine phytoplankton to the global autotrophic biomass is less than 1%, but it accounts for nearly 50% of primary production, driving material cycling and energy flow within marine ecosystems, thus playing a crucial regulatory role in the Earth system (McClain 2009). The vertical distribution of subsurface marine phytoplankton not only contains crucial information for marine ecology but also provides crucial information related to the optical properties of water bodies relevant to remote sensing. Oceanic lidar remote sensing inversion modeling techniques are based on the fundamental technology for accurately measuring profiled carbon parameters (chlorophyll, particulate organic carbon, primary productivity, etc.) through laser penetration of water.

The lidar equation provides a functional relationship between the lidar echo signal and the optical properties of seawater. Therefore, by solving the lidar equation based on the echo signal detected by the lidar, information about the properties of the seawater can be obtained. Besides the power of the lidar echo signal received by the photodetector of the laser radar system, it is known from the lidar equation that the equations also have two unknowns: the seawater attenuation coefficient α(z) and the seawater backscattering coefficient β(z). Solving a problem with two unknowns in a single equation requires determining a relationship between α(z) and β(z), which is the difficulty in algorithm inversion. Currently, the commonly used inversion algorithms mainly include the Collis slope method (Collis 1966; Kunz and de Leeuw 1993; Collis and Russell 2005), the Klett lidar ratio method (Klett 1981, 1985), and Fernald lidar ratio method (Fernald 1984; Ben-David et al. 2007; Tao et al. 2008).

Collis has assumed uniformity to address the issue of a system with two unknowns in a lidar equation, which posits that the seawater under consideration is homogeneous. Consequently, the backscattering coefficient of the seawater remains constant without varying with the depth. Under this assumption, the attenuation coefficient of the water can be determined using a straightforward method based on the slope. The slope method is simple and convenient for solving the problem. However, this method assumes that the sea water is a homogeneous medium, which is difficult to achieve in real-world situations. Thus, this method is often limited, and achieving high accuracy in practice is difficult. The Klett method provides the total attenuation coefficient of the water. Since the intensity of the Mie scattering signal is proportional to the wavelength to the power of (1–2), and the intensity of the Rayleigh scattering signal is proportional to the wavelength to the power of 4, when the wavelength is longer, or the particle concentration is higher; then, the Mie scattering signal dominates the water echo signal, whereas the Rayleigh scattering signal is relatively weak and can be ignored. In cases where only a single component needs to be considered, the Klett method is the most effective, which overcomes the limitations of uniform water conditions. It can invert relatively precisely, even with roughly estimated values, if the water transmission rate near the boundary height is small. Therefore, this algorithm is widely used and successful in practice. The interaction between pure water molecules and particles influences the propagation of laser in water. Therefore, both backscattering and attenuation in the water section of the lidar equation should consist of two components: scattering and attenuation by pure water molecules and scattering and attenuation by particles. The Fernald method separates the scattering by pure water molecules and particle scattering in the lidar equation. Determining the lidar ratio between pure water molecules and particulate matter is crucial when using the Fernald algorithm to solve the lidar equation. The lidar ratio Sw of pure water is a constant that can be directly calculated. The lidar ratio Sp of particulate matter is the primary source of inversion error, and it is related to the refractive index, particle size distribution, shape, and other parameters of the particulate matter. It varies greatly in different sea water, making it difficult to estimate the lidar ratio of particulate matter.

Overall, in traditional inversion modeling methods for oceanic lidar, the inversion results are limited by the ill-posed equation solution (one measurement and two unknowns), requiring the assumption of a lidar ratio to the initial value, i.e., the ratio of attenuation coefficient to backscatter coefficient (Churnside and Marchbanks 2017). However, measuring and estimating the lidar ratio for sea water particles is a complex and laborious process. The conversion relationship between the inverted physical quantity of the lidar and the measurement value of the oceanic optical instrument is unclear, making it difficult to directly measure the lidar ratio through the optical instrument. Furthermore, under the influence of the properties of sea water, the lidar ratio fluctuates according to the particle types and sizes of sea water. Also, there are only a few research reports on the measurement and estimation of sea water lidar ratio due to the complexity and unpredictability of the lidar ratio measurement.

A straightforward perturbation retrieval method has recently been utilized (Churnside et al. 2018, 2022; Churnside and Marchbanks 2017; Liu et al. 2018a, b). If the changes in the depth profile are not overly extreme, they can be represented as the sum of their depth-averaged values and a depth-dependent perturbation. Under this assumption, the integral of the attenuation perturbation is negligible and thus can therefore be ignored. This approximation makes it possible to determine the depth profile of scattering. The fundamental assumptions made in these studies were that the background scattering remains constant vertically or over a certain horizontal distance, and any variations observed are considered perturbations from this constant scattering. Moving forward, we assumed that the optical parameters can be separated into two components: a fixed part that does not vary with depth and a variable part. The retrieval method commences by fitting a curve to the logarithmic signal obtained from the lidar. Subsequently, the background scattering signal, representing the non-varying part, is subtracted from the curve, with the subtraction value indicating the signal corresponding to the varying part. Notably, the errors associated with this retrieval technique mainly arise due to the assumption of neglecting the varying part of the attenuation and lidar calibration factor. In this context, Chen et al. proposed a novel hybrid iterative Klett-Perturbation inversion method (K-P), an advanced variation of the simple perturbation approach. Thus, the accuracy of the inversion of the attenuation coefficient is significantly improved without the need for a uniform assumption of the attenuation coefficient (Chen et al. 2021a, b; Chen 2022; Chen et al. 2022; Zhang and Chen 2022). Furthermore, the method uses a straightforward and viable approach to calibrate oceanic LiDAR by comparing lidar backscatter with calculated scatter based on iterative bio-optical models. Unlike existing calibration methods for aerosol lidar, this approach considers geometric losses and attenuation at the atmosphere-sea interface. Furthermore, the iterative retrieval method using bio-optical models can provide more accurate parameters for the Klett algorithm (Liu et al. 2020a, b).

4 Airborne lidar applications

Airborne lidar and associated instruments have been applied in a range of ocean-related research, including marine biology (phytoplankton, zooplankton, fish), marine non-biology (plastics), and physical and optical properties of the ocean.

4.1 Phytoplankton in subsurface

Phytoplankton is the main productivity of the ocean, and satellite-based radiometers have traditionally been used to monitor its presence. With similar sensors and atmospheric correction for different algorithms, low-flying aircraft can achieve similar ocean color measurements and provide finer spatial resolution (Churnside and Wilson 2008). Montes-Hugo et al. (2010, 2011) studied the effect of phytoplankton inhomogeneous vertical distribution on ocean color measurements using lidar profiles. Phytoplankton is also the cause of harmful algal blooms. In August 2014, Moore examined the western basin of Lake Erie with airborne lidar and indicated the vertical structure of a cyanobacteria bloom of the overall cyanobacteria population was influenced mainly by wind speed, followed by solar heating of surface waters (Moore et al. 2019). Churnside stated that lidar data is available for demonstrating the formation of a wide distribution of thin (> 3 m) plankton layers under various conditions, such as upwelling, fresh-water influx, warm-core eddies, and melting sea ice (Churnside and Donaghay 2009; Churnside and Marchbanks 2015; Churnside et al. 2020). These thin layers concentrate the chemical and biological substances, profoundly influencing primary and secondary productivity. The cutting-edge dual-wavelength blue-green lidar, innovatively crafted by the Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, has been utilized for multiple airborne lidar experiments in the sea areas of Linghsui and Sanya Bay in the South China Sea. The proposed K-P model by Chen et al. was used to invert new blue lidar data, effectively revealing the subsurface chlorophyll profile structure up to 80 m underwater for the first time. Based on these findings, the driving mechanisms of monsoons, water depth, and sea surface temperature on the spatiotemporal characteristics of subsurface phytoplankton in the coastal waters of the South China Sea were further elucidated. Figure 4 presents the subsurface phytoplankton vertical structure up to 80 m for the first time (Liu et al. 2018a, b; Chen et al. 2021a, b). Chen et al. also investigated the applicability of this technology for identifying subsurface plankton layers in the inland water, such as the Qiandao Lake. The in-situ measured subsurface chlorophyll maximum layer and phycocyanin maximum layer showed good consistency. The primary results showed that it is effective to retrieve the subsurface phytoplankton based on the airborne lidar system and the retrieval method, and the primary results of the subsurface phytoplankton vertical structure have been illustrated in Fig. 5 (Chen 2022).

Fig. 4
figure 4

Subsurface phytoplankton vertical structure up to 80 m underwater revealed by airborne lidar in the Linshui Sea for the first time. Redrawn and modified from (Chen et al. 2021a, b)

Full size image
Fig. 5
figure 5

Subsurface phytoplankton vertical structure revealed by airborne lidar in the Sanya Bay. Redrawn and modified from (Liu et al. 2018a, b)

Full size image

4.2 Zooplankton fish and mammals

In addition to the study of phytoplankton, Churnside and Thorne (2005) systematically surveyed copepods of the genus Neocalanus genus in Prince William Sound, Alaska. In order to remove the return from low-level scatters, they applied a threshold to lidar data. Furthermore, because the lidar target intensity of plankton has not been directly measured, they changed the threshold level and compared the results with the acoustic results in the eight areas. They concluded that a threshold level of 2.75 is considered the best protocol relative to the background scattering level. At this threshold, the correlation between the lidar and echo sounder results is 0.78.

National Oceanic and Atmospheric Administration (NOAA) lidar has been used more frequently to survey fish than zooplankton (Roddewig et al. 2017; Roddewig et al. 2018; Vannoy et al. 2021; Scofield et al. 2021). The amount of fishery resources is an important issue in fisheries management, and traditional methods of resource assessment include direct sampling, sonar surveys, and aerial surveys. However, these methods have certain limitations, such as the most serious limitations of ground-based techniques like sampling and sonar, which are slow vessel speed, small survey area, and high cost. Aerial surveys can cover a larger area, but the data are less reliable. Lidar has been proved to be most effective in surveying fish that live near the surface. Those fishes mainly include menhaden (Churnside et al. 2010), sardines (Churnside and Wilson 2004; Carrera et al. 2006; Reese et al. 2011), mackerel (Carrera et al. 2006; Churnside et al. 2009a, b), salmon (Churnside and Wilson 2004), mullet (Churnside et al. 2003), capelin (Brown et al. 2002), anchovies (Churnside et al. 2001), herring (Churnside et al. 2011), flying fish (Churnside et al. 2016), trout (Roddewig et al. 2020), and jellyfish (Churnside et al. 2015a, b). Lidar can distinguish between large fish (tuna and salmon) and small fish (sardines and herring). However, lidar signals alone do not allow a clear identification of fish species, and other clues must be considered.

In addition to fish in the surface layer of the ocean, large marine mammals are also directly detected by lidar. Churnside observed a foraging activity in the Bering Sea: 52 humpback whales and thousands of seabirds were observed in a small area. This aggregation is because a dense layer of plankton is on the surface, attempting to avoid predation from the herring school below (Churnside et al. 2011). During an aerial survey near Kodiak Island, Alaska, Churnside et al. (2009a) observed evidence of the whales in the thermal footprints. Generally, aerial surveys of whales and other marine mammals use visual and visible imagery of expert observers; however, visual surveys can only be conducted during daylight hours. Nevertheless, even if the animals cannot, whales can be seen in infrared imagery. Churnside studied the lidar return from bubbles in the ocean, demonstrating that the return will be proportional to the bubble void fraction, independent of the bubble size distribution (Churnside 2010). Thus, an airborne lidar can be used to measure whale vocalizations, which affect bubble size (Churnside et al. 2015a, b). Churnside used thin plankton layers at the pycnocline (Churnside and Donaghay 2009) as tracers of the pycnocline to measure abrupt changes in depth and internal waves of oceanic structure in coastal waters (Churnside et al. 2012) and open ocean (Churnside and Ostrovsky 2005).

4.3 Physical and optical properties of the ocean

Several studies of the physical properties of the ocean have been conducted using lidar and related instrumentation (Churnside et al. 2017, 2018; Churnside and Marchbanks 2017; Roddewig et al. 2020). The way of lidar return depends on the optical properties of water has been studied using comparisons with in-situ optical measurements (Churnside et al. 1998, 2017; Lee et al. 2013), direct measurements of polarization components (Churnside et al. 1998, 2012; Churnside 2008) and theoretical studies (Mitra and Churnside, 1999; Churnside 2008). As long as orthogonal polarization is utilized for plankton layer detection, there appears to be no contrast difference between linear and circular polarization (Mitra and Churnside, 1999). The statistics of variable scattering in the ocean were also investigated using subsurface lidar return (Churnside and Donaghay, 2009). Chen et al. used a lidar system to procure a comprehensive understanding of the vertical distribution of optical properties in contrasting sea regions, spanning from the East China Sea (ESC) to the South China Sea (SCS) (Chen 2022; Chen et al. 2022; Zhang and Chen 2022). This large-scale observational endeavor stretched across over 3700 km, with a collection of over 74000 lidar profiles captured between September 5 and September 15, 2020, diminishing from the turbid waters of the ESC to the SCS.

Overall, the traditional method for detecting the vertical structure of optical properties and chlorophyll-a involves shipboard discrete observations or Biogeochemical-Argo profiling floats, which take considerable time to cover a limited area. Compared with traditional methods, lidar has the advantages of being large-scale and long-term, and no disturbance is associated with touching measurements. It would be a good complement to passive satellite remote sensing and discrete in-situ observations, allowing us to improve the estimation of phytoplankton primary productivity and carbon stocks/fluxes as well as to understand the temporal and spatial variation characteristics of IOPs and phytoplankton (Chen et al. 2022).

4.4 Internal waves

Observation of the internal waves is attributed to the transformation of energy from barotropic tides into internal waves through an interaction with the shelf break or other bottom features. A NOAA airborne pulsed fish lidar has been employed to observe large, nonlinear isolated internal wave complexes propagating along the flight axis in the Gulf of Alaska. Characterized by internal waves, the obvious scattering layer in the ocean subsurface has moved up and down with the vertical motion of the waves. Using airborne lidar to detect nonlinear internal waves in the sea is highly possible because the plankton layer would have strong oscillations in the depth direction. The reason for the generation of isolated internal wave groups may be the nonlinear offshore propagation of strong internal waves generated by the interaction between tides and the continental shelf (Churnside and Ostrovsky 2005). Churnside has illustrated examples of linear and nonlinear internal waves perturbing the plankton layer and nonlinear isolated internal wave groups propagating along the flight axis (Churnside et al. 2009a, b; Dolin and Dolina 2020). The effects of internal waves on the thickness of the scattering layer have two types according to the vertical displacement. One is that the wavy displacement in the depth of the scattering layer is much smaller than the average depth. Another is that the wave-driven displacement in its layer depth is equivalent to the layer depth. The finding of the observation campaign could be concluded that the plankton layer is related to density gradient, the airborne lidar can easily detect nonlinear internal waves in shallow strait bay, and stratified turbulence leads to spectral dependence, which has a huge impact on lidar backscattering signals (Reineman et al. 2009; Churnside et al. 2012).

Liu et al. (1985) analyzed the evolution of nonlinear internal waves in the Sulu Sea, presented a theoretical model for generating strong nonlinear waves in the thin upper layer above an infinitely deep lower fluid, proved the effectiveness of the nonlinear wave theory, and provided a theoretical foundation for lidar to measure nonlinear internal waves.

4.5 Coastal zone and bottom mapping

Airborne lidar is a technology that uses laser pulses to measure the elevation of the seafloor in shallow waters. Coastal mapping and nautical charting are important for navigation, environmental management, and disaster response. However, it is difficult for shipborne sound navigation and ranging systems to charting the shallow-water areas (Allouis et al. 2010). Airborne lidar bathymetry systems (ALBs) can efficiently provide high-accuracy and high-density bathymetric datasets in non-navigable areas. Owing to superior performance in acquiring the spatial position, ALBs have been widely employed for the seamless topo-bathymetric mapping of shallow-water areas such as near island reefs. Wozencraft and Millar (2005) introduce integrated lidar bathymetry technologies for near-shore mapping and charting. An adaptive depth extraction method based on airborne oceanic lidar is presented by Liu et al. (2018a, b). The rapid development of lidar sensors and software can also be justified by the continuous reviews published by GIM International (Lemmens 2007). With the rapid growth in both the demand and supply, a wide array of bathymetry lidar system applications can be found in various review papers regarding forestry modeling and analysis (Wulder et al. 2012), habitat ecology (Bradbury et al. 2005), landslide investigation, 3D building modeling (Wang 2013), road extraction, and snow depth measurement (Deems et al. 2013).

5 The way ahead

5.1 Airborne high spectral resolution lidar

High spectral resolution lidar (HSRL) is a promising technology that can potentially advance oceanic lidar studies (Hair et al. 2016). HSRL offers the advantage of measuring the spectral characteristics of backscattered light, enabling the retrieval of additional information about suspended particles and other oceanic constituents (Churnside et al. 2018). By utilizing HSRL in oceanic lidar systems, researchers can gain valuable insights into the vertical distribution of aerosols and their impact on oceanic processes. Aerosols play a crucial role in modulating the Earth’s climate system and have significant implications for the exchange of heat, moisture, and energy between the atmosphere and the ocean. Understanding the characteristics and behavior of aerosols can provide valuable information about ocean dynamics, air-sea interactions, and the overall climate system (Burton et al. 2013; Dawson et al. 2020). In addition to the aerosol studies, oceanic lidar systems equipped with HSRL channels can potentially improve the accuracy and precision of measurements related to water properties (Zhou et al. 2022). By analyzing the spectral signatures of backscattered light, it becomes possible to retrieve information about water compositions, including dissolved organic matter, suspended particles, and phytoplankton (Müller et al. 2014; Liu et al. 2016). This can greatly enhance our understanding of marine ecosystems, biogeochemical cycles, and the overall health of the oceans (Schulien et al. 2017). Notably, the development and implementation of HSRL in oceanic lidar systems have certain challenges. HSRL requires sophisticated instrumentation and data processing techniques to retrieve the spectral information from the lidar signals. Furthermore, the interpretation of HSRL data in the context of oceanic studies requires further research and validation (Ferrare et al. 2023). While HSRL holds promise for the future of ocean lidar systems, it is just one of the many evolving technologies in the field. Other advancements, such as improved laser technology, higher pulse repetition rates, and enhanced data processing algorithms, are also shaping the future of oceanic lidar (Zhou et al. 2019). Combining these technologies and ongoing research will contribute to the continued advancement of oceanic lidar systems, enabling a deeper understanding of the marine environment and its interactions with the atmosphere (Ottaviani et al. 2018).

5.2 Unmanned aerial vehicle lidar

UAVs have emerged as promising platforms for remote sensing of ocean environmental parameters and are expected to play a significant role in the future (Yang et al. 2015). UAVs have demonstrated their potential in various oceanic research applications, such as coastal mapping, marine plankton distribution surveys, water quality monitoring, and marine ecosystem studies (Aniceto et al. 2018).

UAVs offer several advantages for oceanic remote sensing compared to traditional platforms like satellites or manned aircraft (Hodgson et al. 2013; Colefax et al. 2018). First, UAVs can be deployed in targeted areas and navigate at low altitudes, allowing for high-resolution data collection and access to remote or challenging locations that may be inaccessible to other platforms. This flexibility enables researchers to obtain data with greater spatial and temporal coverage (Koh and Wich 2012). Second, UAVs are generally more cost-effective than manned aircraft or satellite missions. They have lower operational and maintenance costs, making them a viable option for researchers with limited budgets. Additionally, UAVs can be easily deployed and operated, reducing logistical complexities (Anderson and Gaston, 2013). Third, UAVs can have sensors and instruments tailored to specific research needs. This flexibility allows for integrating different remote sensing technologies, including lidar, multispectral, hyperspectral imaging, and thermal sensors. Researchers can optimize sensor configurations based on the targeted oceanic parameters and research objectives (Colomina and Molina 2014; Linchant et al. 2015). A lightweight UAV-borne lidar bathymeter is also developed to detect shallow-water depth and the underwater target (Wang et al. 2022; Huang et al. 2023).

5.3 Blue carbon monitoring using airborne or spaceborne lidar

The vertical distribution of phytoplankton in marine waters is typically heterogeneous. The primary productivity of subsurface phytoplankton contributes significantly to the entire water column and exhibits the highest proportion of new productivity. This new productivity reflects the ocean’s ability to absorb CO2 from the atmosphere in the euphotic zone, which is paramount for estimating marine primary productivity and understanding the global carbon cycle. However, the lack of long-term observational data has severely hindered our understanding of the distribution patterns, influencing factors, and formation mechanisms of subsurface phytoplankton profile structures. Recent studies have found that incorporating profile structure information in estimating primary productivity can improve accuracy by more than 50% (Barbini et al. 2005; Schulien et al. 2017). This revelation highlights the non-negligible contribution of underwater subsurface phytoplankton carbon sequestration parameters throughout the upper ocean water column.

Given the advantages of oceanic lidar in profiling and continuous observation during both day and night, it will undoubtedly play a significant role in future blue carbon monitoring. The conventional passive ocean color remote sensing techniques fall short in providing nocturnal data. Research on diurnal variations in global surface-level chlorophyll, particulate organic carbon (POC), pCO2, and ocean-atmosphere carbon dioxide flux (C-Flux) reconstructions is limited. However, as an active remote sensing technology, the lidar triumphs over the limitations imposed by solar irradiance, enabling night-time and polar region observations and yielding high spatiotemporal resolution vertical information of the subsurface oceanic realm. Hence, it is the most promising marine remote sensing technique in contemporary times (Jamet et al. 2019).

Researchers have recently explored the potential of the CALIPSO cloud and aerosol lidar meteorological satellite, originally not intended for marine applications. These endeavors have uncovered the prospects of backscattering remote sensing of subsurface particles, offering novel means for global ocean ecosystem observations (Behrenfeld et al. 2019; Lu et al. 2020). Lu et al. used the ICESat-2 satellite lidar measurements for the two-dimensional distributions of upper ocean phytoplankton properties. The spring phytoplankton blooms extending about 230 km horizontally from dense packs of ice near Antarctic marginal ice zones and 15 m vertically below the ocean surface are observed from space for the first time (Lu et al. 2020, 2021). Recently, Chen et al. have developed a novel lidar remote sensing methodology for diurnal variations in global and polar chlorophyll and ocean-atmosphere carbon flux, utilizing satellite-based lidar (CALIPSO) remote sensing data and a physics-driven deep learning model. This methodology relies on a physics-driven deep neural network (DNN) model, employing CALIPSO total attenuated backscatter ratio and water column-integrated backscattering coefficient (bbp) as inputs, combined with MODIS daytime bbp and POC products for training and model establishment. Subsequently, the model retrieves bbp, chlorophyll, and POC products from CALIPSO data in regions not covered by MODIS, such as night-time and polar areas. Unlike previous physics-based approaches (Behrenfeld et al. 2019), this method does not require the diffuse attenuation coefficient and avoids assumptions about the relationship between bbp and 180° backscattering coefficient β(π). Preliminary results indicate that the CALIPSO products derived using the physics-driven DNN method align more closely with traditional ocean color remote sensing daytime products than conventional physics-based methods (Zhang et al. 2023a, b).

Oceanic lidar, with its depth resolution capabilities, enables continuous day-and-night observations of the profile structure of phytoplankton in global and polar regions. This technology aids in improving the accuracy of global phytoplankton primary productivity and organic carbon estimation. Therefore, developing oceanic lidar detection techniques is necessary to transition from remote sensing of carbon parameters solely at the sea surface and daytime to remote sensing of carbon parameters within the subsurface profile structure and continuous day-and-night observations.

6 Conclusions

Large-scale optical characteristics of the upper ocean and biological phenomena in the subsurface have been well observed and researched. However, the airborne lidar system can provide a high spatial and temporal resolution view of plankton distribution and activities. Spurred by the improved laser technique and advanced photonics sensors, airborne lidar systems have performed important roles in the remote sensing of the oceanic parameters, which could build a path for refining the quantification of biogeochemical importance. This report attempts to fill the gap in airborne oceanic lidar reviews by focusing on the oceanic parameters.

However, regarding operational observation of the oceanic environment, some issues should be considered, including the quantitative and automatic or semi-automatic lidar data processing, enhancing system stability, and higher detection performance to support long-term operational lidar. First, the oceanic parameters retrieval algorithm of lidar data processing should put great effort into providing accurate information in high depth. Second, the stability of the lidar systems should be enhanced to support long-term observation based on unmanned aerial systems with longer endurance. Third, it is critical to consider improving subsurface (up to 100 m) measurements for oceanic environment parameters.

Robust and cost-effective lidar systems networks are essential to achieve broader spatial coverage of the oceanic environment and provide a more comprehensive understanding of oceanic processes and features. While having operational lidar networks and maintaining high-quality data collections are necessary for operational data assimilations, further developments of assimilation methods to effectively use highly temporally and vertically resolved lidar measurements are still needed. More efforts are required to encourage and develop intelligent data processing tools compatible with multi-platform and open access standards.