Chromophoric dissolved organic matter (CDOM) which represents the colored fraction of dissolved organic pool, can strongly influence ocean optical properties, remotely sensed spectra and biogeochemical processes [1-3]. The laser-induced fluorescence (LIF) technique is a well-known analytical technique for rapid water environment monitoring, which is based on the measurements of laser-induced water emission spectrum, to obtain qualitative and quantitative information about the in-situ fluorescent constituents[4, 5]. Compared with traditional measuring methods for CDOM monitoring, LIF technique has the advantage of rapidly acquiring high-resolution in situ profiles . Spectral deconvolution is one of the key components in fluorescence data processing. In order to analyze fluorescence spectral complexity by the overlap of various fluorophore spectra, kinds of spectral deconvolution methods were used [7-9]. For instance, radial basis function networks was developed by Li in 2005 to de-convolute overlap spectral components, and the Pearson’s IV function(s) was introduced by Chekalyuk and Alexander in 2008. Yet no mature general method has been applied well for all kinds of spectral shape until now. The Pearl River is a complex river network under the influence of heavy urbanization and industrialization. Studying the utility of LIF technique in the Pearl River Estuary is, therefore, of significance. In this paper, we described a dual-wavelength lidar fluorosensor system for fast diagnosis of chromophoric dissolved matter (CDOM) in water in the Pearl River estuary. In order to assess the performance of this new system, we compare the measurements taken by the instrument to those by laboratory absorption spectrophotometer. Applications of the new system for water monitoring studies are illustrated by in situ measurements obtained in the Pearl River Estuary.
INSTURMENT AND METHOD
The locations of sampling sites are plotted in Fig. 1. A total of 18 discrete samples were measured from the surface water in the Pearl River Estuary in December 2013. Temperature, turbidity, PH and salinity were obtained by a multiparameter water quality monitor (Manta, Eureka Inc.) during the cruise. In-situ fluorescence measurements on unfiltered samples were obtained using the laser fluorometer. Absorption measurements were analyzed in a spectrophotometer after the cruises, approximately three days after sampling.
A portable analysis instrument for fluorescence spectroscopy induced by lasers is designed in this paper. The LIF optical system is established based on orthogonal Czerny-Turner optical configuration. It consists of an excitation source module, a sample holder module, and a detection module (Fig. 2). Two micro violet laser with 355 nm and 532 nm wavelength (MM-405-100, Boson Tech) are selected as excitation source, and a hyper-spectral CCD camera (USB4000- FL, Ocean Optics) is integrated to record complete visible LIF emission spectra with a range from 360nm to 1000nm, with high spectral resolution (0.2nm). The laser is with high pulsed repetition frequency rate (greater than 10 kHz) and low power consumption (about 10 w). Two reflecting mirrors (74-msp, Ocean Optics) are used to enhance the fluorescence signals. The coupling facilities of the LIF systems are highly integrated components with no need for optical path adjusting. During the measurements, the output of the laser is focused into a quartz sample cell, then the emitted fluorescence light is collected by a collection quartz lenses onto the entrance slits of a spectrometer, which is interfaced to a computer through USB port. Ambient and background noises is subtracted from raw spectra and water Raman scattering is used to correct laser fluorometer data for effects of water optical attenuation. The overlapping fluorescence spectra of water Raman scattering and CDOM were separated with fitting bi-Gaussian of the least squares method. The LIF spectral integration time is typically preset 0.1 to 1 s, and the number of acquisitions is typically preset 5-25, in order to average the fluorescence spectra. Each sample measurement typically costs 30 to 50s. A diagram of the fluorescence lidar instrument is presented in Fig. 3.
LIF spectra deconvolution
Spectra deconvolution methods often use a non-linear optimization algorithm to decompose a complex overlapping-peak signal into its component parts. The LIF spectra is recorded as a multi-dimensional matrix of intensity and spectral parameters. Spectroscopy peaks are frequently observed to be asymmetric in which cases the peak maximum loses its significance and the calculation of peak area is also complicate. The bi-Gaussian peak function was used for LIF spectral deconvolution to retrieve asymmetric approximations of the basic spectral components in this study. LIF spectra were fitted by the following bio-optical components: water Raman scattering (WRS), chlorophyll-a fluorescent (Chl-a), and CDOM fluorescent (FDOM). All components are described in the following way by bi-Gaussian peak function:
Here, y0, xc, H, w1, w2 are the parameters that define the base line, the location of the peak, the peak height, the width of the half Gaussian function to the left and the width of the half Gaussian function to the right.
The spectral curve of absorption and fluorescence are shown in figure 4. The absorption coefficient range from 0.44 to 1.21 m-1, and the fluorescence intensity in Raman unit(RU) range from 0.4 to 1.6 with 532 nm laser stimulation and range from 0.77 to 1.77 with 355 nm laser stimulation. It illustrates that the peak at 470 nm corresponds to the Raman scattering peak in the water, the peak nearby 500 nm corresponds to the CDOM fluorescence peak when using the 355 nm laser excitation, and the peak at 650 nm corresponds to the Raman scattering peak in the water, the peak nearby 580 nm corresponds to the CDOM fluorescence peak when using the 532 nm laser excitation Up to 2-5 orders of magnitude of variability in the CDOM fluorescence intensity were observed due to significant variation in CDOM content in different water types. It reveals the fluorescence spectral variability in diverse water types.
An example of bi-Gaussian deconvolution for fluorescence emission spectrum measured by a 405 nm laser fluorometer is presented in Fig. 5. The spectral curve presents several peaks and should reveal different spectral constituents. Because the visible LIF emission is mostly caused by CDOM and different phytoplankton pigments, so the emission spectra by the laser fluorometer can be regarded as an overlapping spectra integrated by the water Raman band, the CDOM constituent band and the different pigments bands. In the case of overlapping spectral features, a multiple bi-Gaussian deconvolution was applied to indicate the integration band for each deconvoluted peak. The peak nearby 470 nm corresponds to Raman scattering in water, the peak nearby 508 nm corresponds to CDOM constituent, and the peak nearby 685 nm corresponds to Chl-a pigment. It reveals the capacity of the LIF technique for monitoring various water biochemistry constituents.
Figure 6(a) displays the linear correlation of R2 = 0.97 between laser fluorescence intensity and ag(355) with the 355 nm stimulation, and figure 6(b) displays the linear correlation of R2 = 0.94 between laser fluorescence intensity and ag(355) with the 532 nm stimulation. The correlation between fluorescence intensity and ag(355) demonstrates that LIF system provides effective and significant result for CDOM assessment. Additional consideration and work should be carried out when measuring higher amounts of suspended solids in the water.
The main goal of this study was to test applicability of the LIF system in determining CDOM levels in Pearl River estuary, and to verify and establish correlations between CDOM fluorescence and absorption laboratory analysis. The significant correlation between fluorescence by the technique and absorption by laboratory spectrophotometer indicates the good capacity of the LIF technique for charactering CDOM. It should be a good tool for research and observations in various waters at high-resolution, high-frequency and longer terms. Lot of work still needs to be carried out to fully study the characteristics of CDOM in estuaries.
A. Bricaud, A. Morel, and L. Prieur, “Absorption by dissolved organic matter of the sea (yellow substance) in the UV and visible domains,” Limnol. Oceanogr 26, 43–53 (1981)Google Scholar
Carder, K.L., et al., Reflectance model for quantifying chlorophyll a in the presence of productivity degradation products. Journal of Geophysical Research: Oceans (1978-2012), 1991. 96(C11): p. 20599–20611.Google Scholar
Arrigo, K.R. and C.W. Brown, Impact of chromophoric dissolved organic matter on UV inhibition of primary productivity in the sea. Marine ecology progress series. Oldendorf, 1996. 140(1): p. 207–216.Google Scholar
Chen, P., D. Pan, and Z. Mao, Fluorescence measured using a field-portable laser fluorometer as a proxy for CDOM absorption. Estuarine, Coastal and Shelf Science, 2014. 146(0): p. 33–41.Google Scholar
Chen, P., D. Pan, and Z. Mao, Development of a portable laser-induced fluorescence system used for in situ measurements of dissolved organic matter. Optics & Laser Technology, 2014. 64(0): p. 213–219.Google Scholar
Rogers, S.R., et al., Airborne Laser-Induced Fluorescence (LIF) Light Detection and Ranging (LiDAR) for the Quantification of Dissolved Organic Matter Concentration in Natural Waters. Estuaries and coasts, 2012: p. 1 - 17.Google Scholar
Zhao Nanjing, L.W., Cui Zhicheng et al, Analysis the characters of dissolved organic matter in water using spectral fluorescent signature. Acta Optica Sinica, 2005. 25(5): p. 687–690.Google Scholar
Xue Zechun, L.L., Liu Ying, Application of wavelet transformation and Gaussian fitting in resolution of Multi-component overlapping spectra analyzing. Journal of Chongqing University of Arts and Sciences, 2011. 30(5): p. 44–47.Google Scholar
Babichenko, S., et al. Airborne surveillance of water basins with hyperspectral FLS-LiDAR in Remote Sensing. 2010. International Society for Optics and Photonics.Google Scholar
Li HongBin, L.W., Zhang YuJun, A method of characteristic LIF spectral signatures separation based on rddial basis function networks. ACTA PHYSICA SINICA, 2005. 54(9): p. 4451–4457.Google Scholar
Chekalyuk, A.M. and M.A. Hafez, Advanced laser fluorometry of natural aquatic environments. Limnology and Oceanography: Methods, 2008. 6: p. 591.Google Scholar
Buys, T. and K. De Clerk, Bi-Gaussian fitting of skewed peaks. Analytical Chemistry, 1972. 44(7): p. 12731275.Google Scholar