OCEANIC LIDAR: THEORY AND EXPERIMENT

Study on the upper ocean is of great significance to the global climate change and carbon cycle. Lidar can be used to effectively detect depthresolved optical properties of the ocean. However, both theory and experiment of oceanic lidar are limited by complex multiple scattering. Several progresses by Zhejiang University will be illustrated in this paper: 1) a polarized lidar system was developed, and a Monte Carlo model and a radiative transfer model were established (Zhou, et al. remote sensing, 2019; Zhou, et al. Journal of remote sensing, 2019; Xu, et al. and Liu, et al. Journal of remote sensing, 2019); 2) Cross validations are demonstrated to verify the availability of the lidar system and models (Liu, et al. IEEE TGRS, 2019); 3) phase function effects on backscatter and attenuation are studied considering multiple scattering, respectively (Liu, et al. Optics Express, 2019). Oceanic lidar is proven to have great potential in marine studies.


INTRODUCTION
Study on the upper ocean is of great significance to the global climate change and carbon cycle. The ocean color remote sensing based on the airplane or satellite, like SeaWiFS, is able to collect the global data over a long term efficiently. Nevertheless, the limited information about the depth and the dependence on the natural light partly restrict its applications. So far, the oceanographic lidar, one of the active remote sensing methods, has been employed in detecting fisheries, phytoplankton layers and internal waves, etc., in the upper ocean [1][2][3].
In general, interpretation of lidar signals is not difficult under the well-known single-scattering approximation. However, in dense media, such as clouds, seawater, etc., the preceding and following events are accompanied by strong multiple scattering [3]. A portion of photons that are lost in the scattering events could eventually contribute to the lidar signals through multiple scattering. As a result, large errors could be introduced into the single-scattering approximation. Therefore, it is important to quantify and analyze lidar signals with multiple scattering.
Several progresses by Zhejiang University will be illustrated in this paper: 1) a polarized lidar system was developed, and a Monte Carlo model and a radiative transfer model were established; 2) Cross validations are demonstrated to verify the availability of the lidar system and models; 3) phase function effects on backscatter and attenuation are studied considering multiple scattering, respectively.

The lidar system
The developed shipborne oceanic lidar transmitted a laser pulse into the seawater and detected the lidar returns for the retrieval of the seawater optical properties, as shown in Fig. 1. The transmitter included a frequency-doubling Qswitched Nd:YAG pulsed laser at 532 nm, with a single pulse energy of 5 mJ, a pulse width of 10 ns and a repetition frequency of 10 Hz.

Monte Carlo models
The standard MC and semianalytic MC models are employed to provide MC-simulated results. The basic principle of the MC simulation is to treat photons as classical particles [4] and simulate the trajectories of a number of photons to measure the relevant information. The standard MC algorithm refers to the method described in [4]. The semianalytic MC algorithm refers to the method illustrated in [5], where an analytical estimate is calculated for the possibility of the collection of scattered photons at certain points.
The semianalytic MC algorithm greatly improves the calculation efficiency compared with the standard MC algorithm [6].

Analytical model
The analytical model employed in this paper follows the work of Katsev [7] and Malinka [8]. Seawater has very sharp forward peaks in its phase function that make the probability of scattering into the near-forward directions much larger than that of scattering into the backward hemisphere. If the optical thickness of the media is not too large, it is assumed that the trajectories that contribute to lidar signals primarily consist of single backscattering and small-angle forward multiple scattering on the outgoing and returning legs [7], namely the QSA approximation, which forms the foundation of the analytical model [6].

Cross validation
The lidar signals calculated by the analytical model, semianalytic MC algorithm and standard MC algorithm are shown in Fig. 2. The effects of height, FOV and water type are shown in Fig. 2 (a)-(c), respectively. The semianalytic MC and standard MC algorithms are simplified as "Semi MC" and "Stan MC", respectively, in each legend. To give a more explicit picture, the lidar signals are normalized by their maximums, and the dynamic ranges are set to 4 orders of magnitude. In Fig. 2, the results of the three algorithms agree very well at different conditions [6].   Fig. 3, the analytical model can perfectly match the lidar-measured results at different stations [6,9].

Phase function effects
The effective 180° VSF p π β ′ can be obtained from the simulated signals.  The lidar attenuation coefficient α can be retrieved, as shown in Fig. 5. The sum of the absorption and backscattering coefficients ( ) b a b + (black dashed lines) is used as a reference because α should be always greater than ( ) b a b + under QSA approximation [3,11]. The term α is close to ( ) b a b + for all phase functions at the water surface and increases with depth because of loss during multiple scattering.