Magnitude and quantum yield (η) of sun induced chlorophyll fluorescence are determined in widely varying productive waters with chlorophyll concentrations from 2–200 mg/m3. Fluorescence was estimated using linear fitting of in-situ measured surface reflectance with elastic and inelastic reflectance spectra. Elastic reflectance spectra were obtained from Hydrolight simulations with measured absorption and attenuation spectra as inputs. η is then computed based on a depth integrated fluorescence model and compared with Hydrolight calculation results. Despite the large variability of coastal environments examined the η values are found to vary over a relatively narrow range 0.1%–1% with mean values of 0.33%±0.17%.
©2008 Optical Society of America
The detection of solar induced chlorophyll fluorescence (SICF) obtained by processing ocean color spectra from satellite sensors (e.g. MODIS and MERIS) has been a very powerful tool for monitoring marine phytoplankton on synoptic scales [1–3]. As a signal specific to phytoplankton, SICF provides an alternative means to assess the biomass and primary productivity. Compared with the traditional blue-green algorithms that take advantage of the phytoplankton absorption features in the blue-green spectral region, SICF retrievals are less likely to be contaminated by colored dissolved organic matter (CDOM) absorption or by highly scattering mineral particulates . In addition, atmospheric corrections in the blue band are more difficult than ones in the red band. Therefore, a near infrared (NIR) based algorithm, that retrieves SICF, can potentially provide significant additional information about phytoplankton physiology and improve our understanding of marine ecology on both local and global scales .
Connecting chlorophyll fluorescence data to biomass is complicated by the fact that the effective quantum yield η (the efficiency in the conversion of absorbed to emitted photons) is difficult to quantify  under natural light conditions. Any attempt to use the fluorescence signal as a proxy of chlorophyll concentration requires a-priori estimates of η, which is known to be affected by a number of highly variable environmental factors. At the same time, the phytoplankton abundance and its pigment compositions also have impacts on the reabsorption factor of fluorescence Q*a which is a contributing factor to η. Several models have been developed to parameterize η as a function of solar irradiance, nutrient status and phytoplankton optical properties [7–9]. However, due to the large uncertainties that are difficult to quantify in these complex models, this paper focuses only on the range of variation in η that occurs as a result of the combined inherent complexities of eutrophic coastal environments. Clearly, this type of information about the extent of variation inη in the variety of coastal environments is very useful to help quantify uncertainties in chlorophyll estimates from fluorescence data, and for realistically modeling and estimating contributions of SICF to the remote sensing reflectance.
Although laser spectroscopy can provide improved selectivity and sensitivity for characterization of the fluorescence emission signature of water constituents in real time , their significance is limited because single wavelength nature of the excitation does not model nature sufficiently well. While non-invasive retrieval of SICF η from field passive measurements of spectral radiance have been performed by several groups, these measurements were limited to waters with low chlorophyll concentration (up to 6 mg/m3) and can not be easily generalized to more optically complicated conditions[9, 11, 12]. In a recent study  we found that the effective quantum yield of chlorophyll fluorescence was around 0.5% based on the analysis of the position of NIR peaks in the measured reflectance as a function of chlorophyll concentrations of up to 250 mg/m3. Although this finding was tested on an extensive synthetic dataset using Hydrolight, with highly variable optically active constituents typical of coastal waters, further confirmation is required from field data, which is not readily available.
Since the upwelling radiance is contributed to by both elastic and inelastic scattering, derivation ofη requires accurate determination of the SICF component in the total radiance. In previous studies where chlorophyll concentrations were fairly low, this was simplified because the elastic scattering component (i) is small enough that the non-fluorescent background in fluorescence region can be reasonably approximated by a straight line (fluorescence light height FLH), (ii) is totally ignored for data obtained from deeper water or (iii) is modeled by assuming a power law function for backscattering spectrum [9, 11, 12]. In highly productive waters the situation is more complicated. First the combined effects of water and phytoplankton absorption create a nonlinear elastic reflectance spectrum in the fluorescence band that could become dominant with increasing concentration of chlorophyll and/or non-algal particles . Second the spectral signature of backscattering can be significantly modified by backscattering from the phytoplankton itself in the NIR region . Neglecting either of these factors when estimating SICF from radiance measurement for chlorophyll-rich and optically complicated waters can result in significant errors. In addition to SICF, other inelastic scattering processes including Raman scattering and CDOM fluorescence can affect the total radiance signal . However, the importance of the former is only limited to clear water nearly devoid of chlorophyll  while the latter can be as high as 70% for blackwaters with predominantly high CDOM and virtually no particulates but becomes negligible when scattering particles are present . For highly turbid coastal water conditions encountered in this paper, the contribution of inelastic scattering, other than chlorophyll fluorescence is estimated to be less than 5% of the total reflectance in the blue green region. As a practical matter, therefore, only SICF need to be considered for spectrum fitting purposes.
The present approach takes advantage of hyperspectral field measurements of absorption and attenuation, which are combined with remote sensing reflectance and used to determine SICF and its quantum yield in highly productive coastal waters. In order to fully take into account the impacts of the spectral features of IOP on the reflectance, vertical distributions of high resolution absorption and attenuation spectra are fed into the radiative transfer program (Hydrolight)  to properly characterize the elastic reflectance spectra impacted by the scattering and absorption. A linear least square fitting (LSQ) is then applied to the in-situ remote sensing reflectance to decompose the signal into elastic scattering spectra and the chlorophyll fluorescence spectra to obtain the SICF radiance. With chlorophyll fluorescence quantification and phytoplankton absorption retrieved from absorption field data by LSQ fitting based on bio-optical modeling, theη is derived from in-situ data using a depth integrated fluorescence model. Finally, the statistical information of η has been examined and analyzed for a large variety of productive coastal waters.
2. Instruments and field measurement
In-situ measurements were carried out at 42 stations across the Chesapeake Bay area, Maryland over 10 days from July 11th to 20th in 2005 (CB) and 48 stations at Long Island Sound and its vicinity in summer 2007 from June to September (LIS). Strong algal bloom (brown tide) conditions occurred during both CB and LIS trips.
2.1 Radiance measurements
The upwelling radiance Lu(λ,0-) was collected using a fiber bundle placed just beneath the water surface and connected to a GER 1500 spectroradiometer. The fiber bundle consists of 94 silicon fibers with core diameters of 400µm. The downwelling irradiance above surface Ed(λ,0+) is measured by pointing the same probe bundle to a Spectralon plate (Labsphere PT# SRT-99-020). The GER spectroradiometer had undergone absolute radiometric calibration before and after the cruise to confirm the consistency of instruments response over time. The remote sensing reflectance below surface was obtained from normalization of the upwelling radiance to the downwelling irradiance:
In the CB cruise, the Lu(λ,0-) and Ed(λ,0+) value were also measured in parallel using USB2000 Ocean Optics spectrometers with a fiber probe (a single 400um silicon fiber) below water for radiance and another probe with cosine collector pointed towards the sky on an unshaded mast of boat for irradiance .
2.2 IOP instrument package and their measurements
An in-situ profiling package assembled by WET Labs (Philomath, Oregon) was deployed to record the related water properties. This package consisted of three instruments: an ac-s recording total absorption and attenuation of non-water constituents (anw(λ) and cnw(λ)) at 82 wavelengths nearly equally spaced between 400 nm and 750 nm, giving an average spectral resolution of 4 nm, an ECO-bb9 scattering meter with 7 visible channels for particulate backscattering coefficient bbp and the other two channels for the fluorescence of chlorophyll and CDOM respectively, and a CTD to obtain temperature, salinity and depth data. A filter (Critical Process Filtration Inc.), allowing only the dissolved matter with sizes less than 0.22µm to pass through, was installed at the inlet of the absorption tube of the ac-s meter to measure CDOM absorption. In Long Island trips (LIS), the ac-s recordings were performed with and without the filter. The ac-s instrument was properly calibrated using optically pure water as a reference (Barnstead NANOpure) before and after each cruise in order to track and compensate for any drift. The depth of the profiler was raised and lowered either manually or by automatic winch in order to sample the depth profile of the ocean. The whole system was cleaned with pure water after every day of measurement.
ac-s data were corrected for temperature, salinity and scattering according to manufacturer recommendations. All data were sorted according to depth and binned to no less than 0.1m interval. All the rows containing bad a and c due to bubbles were manually removed. Finally, the particulate backscattering ratio b̃bp was obtained from anw, cnw and bbp data as b̃bp=bbp/(cnw-anw).
2.3 Water Sample analysis
For both CB and LIS cruises water samples were also collected and analyzed for composition of chlorophyll, total suspended solid (TSS) within 6 hours of collection. Chlorophyll concentrations C were obtained using filtration through glass fiber filters (Whatman GF/F) within several hours of sample collection, freezing of filters for at least 24 h, extraction with grinding in 90% buffered acetone, and spectrometric determination with the trichromatic equation . Total suspended solids (TSS) concentrations were determined gravimetrically with tared, pre-ashed glass fiber filters. In CB trip CDOM absorption at 400nm ay0 was determined from absorption measurement of the sample filtrate at two wavelengths 360nm and 440nm.
3. Theoretical background of SICF
3.1 SICF modeling
For nadir viewing the chlorophyll fluorescence radiance emitted at λem from depth z of a homogeneous water body depends on the chlorophyll absorption achl, scalar irradiance at this level Eo(λ,z) (mol.m-2s-1nm-1), the quantum yield η and the total absorption q(λem) at the emission band :
The factor 4π is the result of converting isotropic fluorescence to radiance. a is the total absorption. G(λem) defines the emission spectral distribution which is independent of the excitation wavelength for chlorophyll and is commonly modeled as Gaussian function centered at 685nm with 25nm FWHM [21,22] although more complex spectral fluorescence shape has been found in the laboratory experiments . ex Λ represents the excitation band which is from 400nm to 700nm. Note that the quantum yield here is the so called effective quantum yield  * η=φ·Q*a, which takes into account the reabsorption factor of fluorescence Q*a as well as the quantum yield φ. The dependence of Eo(λ,z) on depth is modeled by an exponential decreasing function:
where K(λ) is the diffuse attenuation coefficient for Eo(λ,z) and Eo(λ,0-) is the scalar irradiance just below surface. The radiance just below surface due to the chlorophyll fluorescence at em λ is simply the integration over the whole water column:
Eo(λ,0-) is transformed from below surface downwelling irradiance Ed(λ,0-) by applying an average geometrical correction factor . To estimate the diffuse attenuation coefficient, we use :
where bb is the backscattering coefficient and θs is the solar zenith angle.
3.2 Hydrolight simulation
In order to verify the relationship between the chlorophyll fluorescence component of the upwelling radiance Lf and the inherent optical properties (IOP) of water as stated in Eq. (4) and (5) a large number of simulations for the water leaving radiance spectra with and without including fluorescence were generated to a 1nm resolution using Hydrolight. A four-component model was used to generate the IOP input for the program, which includes pure water, CDOM, non-algal particulates (NAP) and phytoplankton (Chl). To account for the variability of coastal waters, contributions to the IOP from each component were allowed to vary independently within certain reasonable ranges found in the literature and field data in a manner similar to the IOCCG database construction . More details of the parameters used can be found in . The sun angle was fixed at 30° using clear sky conditions, with wind kept at 5m/s. From these simulations, the magnitude of the Chl fluorescence was determined by subtracting the simulated spectra without fluorescence from that with fluorescence included.
The validity of Eq. (4) and Eq. (5) can be seen by the nearly perfect y=x line match when comparing the modeled fluorescence radiance at 685nm to the Hydrolight simulation results as shown in Fig. 1 with η=0.01 applied. The difference between them is around 3% and might results from errors in estimations of K(λ) and µ̄.
4. Algorithm description
Once the fluorescence model has been checked over realistic conditions, the algorithm to calculate η from in-situ spectral data can be divided into the following 3 steps: 1) estimate the Chl absorption achl from the absorption measurement; 2) determine the component of the radiance signal contributed to by Chl fluorescence Lf from the measured rrs, anw, cnw and Ed(0+) ; 3) compute η using the retrieved parameters of achl, fluorescence radiance Lf and b̃bp.
4.1 Bio-optical modeling and Chl absorption retrieval
The measured absorption of non-water constituents anw was decomposed into absorption due to Chl achl and a non-algal absorption component that includes CDOM and NAP adg using a nonlinear LSQ fitting scheme. The adg typically fits the exponential functions with slope Sdg and amplitude adg0 while achl is modeled by the standard chlorophyll absorption spectral shape (normalized at λ 0) a0chl times an amplitude achl0 
After adg was determined through the fitting, achl is retrieved by subtracting the exponential part from anw. This process allows the spectral shape of achl to differ from the initially assumed a0chl. The fitting residual is defined as a relative root mean square:
where λi is the i-th wavelength channel, M is the number of available wavelengths, and ānw a is the mean value of measured absorption over 82 visible wavelengths.
For stations where the field data of CDOM absorption were available, the fitting based on the exponentially decreasing function was used to obtain the CDOM absorption at λ0=400nm ay0:
ay0 was considered as the direct field truth data and compared with the adg0 retrieved from measured anw later.
At several stations in CB campaigns, extremely high concentrations of non-algal particulate led to saturation of the attenuation measuring instruments at the short wavelength channels. Since a smooth decrease of attenuation with wavelength was always observed in the field data, the unsaturated attenuation segment at long wavelengths was fitted to a power law model: and extrapolated into the short bands.
4.2 Calculating chlorophyll fluorescence from remote sensing reflectance
The elastic part of remote sensing reflectance is related to water IOPs of water by:
where a is the total absorption and backscattering bb is the sum of particulate and water backscattering (bbp and bw); κ is a coefficient depending on the illumination conditions. For productive waters a is the dominant term in the denominator and bw is negligible compared with bbp. Thus assuming a wavelength independent backscattering ratio of particulate b̃bp defined as bbp=b̃bp·bp the spectral dependence of rrs is mainly dependent on bp and a while the amplitude is proportional to b̃bp. In fact radiative transfer simulations confirm that rrs depends linearly on b̃bp for the range of IOP field data encountered. With these observations the spectral shapes of the elastic components of remote sensing reflectance r̂rs were obtained through Hydrolight simulations, without including Chl fluorescence and by inputting the vertical distribution of anw and cnw from the ac-s measurements and a base particulate backscattering ratio b̃bp0 of 1%. Once the shape of the elastic component is estimated, a linear fitting is performed to fit the measured rrs through:
Here, ϕ̂ is the basis vector defining the fluorescence spectral dependence which is a Gaussian function centered at 685nm with FWHM of 25nm; the fitting parameter β is β=b̃bp/b̃bp0 and f is the amplitude for chlorophyll fluorescence. From these two parameters, the amount of surface radiance due to chlorophyll fluorescence at 685nm is simply: Lf(685)=f·Ed(685,0+)·G(685) and the particulate backscattering is bbp=b̃bp·(cnw-anw). We note that using the Hydrolight rather than the analytical remote sensing reflectance model [25, 26] gives a more accurate spectral description of non-fluorescent light by reducing errors in the estimation of κ. A relative root mean square χ 2 is used to quantify the difference between the fitted rrs,fit and the measured rrs reflectance spectrum:
4.3 Determination of chlorophyll fluorescence quantum yield η
Although the depth distribution of anw and cnw is included for the background estimation of the elastic scattering component, the quantum yield of chlorophyll fluorescence η is computed based on a homogeneous water assumption. This is because the total absorption in the fluorescence region is too large for light to penetrate beyond 2–3m of water for most of the stations due to high water absorption and high chlorophyll concentration. In fact the depth profile of the top 5m shows variations in IOP values of less than 15% of the mean values. Therefore only the IOP data closest to the surface were used to compute the value of η. The Ed(λ) just below surface is obtained from the above surface measurement by taking into account a transmittance for air-water interface t ~0.98 : Ed(λ,0-)=tEd(λ,0+). Thus using the fitting parameters of Lf, achl, b̃bp and the measured upwelling radiances Ed(λ,0+) together with the surface value of anw and cnw, η can be determined through Eq. (4) and Eq. (5).
5.1 Variability of coastal water environment
The variations in surface values of optical and water analysis data for each cruise are listed in table 1. Surface chlorophyll concentration C varied from less than 2 mg/m3 to more than 200 mg/m3 under blooming conditions and there are over 10 fold variations in magnitude of ay0 and nearly 40 fold in TSS concentration illustrating a wide variety of productive coastal water conditions. In particular CB is a highly productive turbid estuarine water with a median value of anw at 675nm of 0.67m-1 and particulate scattering at 550nm bp (550) of 8.6m-1 while LIS is characterized by relatively clear productive coastal water with a median value (675) anw of 0.16 m-1 and bp (550) of 2.4m-1.
5.2 Chlorophyll absorption retrieval
To illustrate the efficacy of chlorophyll absorption retrievals, typical fitting results of absorption are shown in Fig. 2. The residuals for absorption fitting χ 1 are less than 0.1 for most cases. We also tested several other spectral shapes of phytoplankton absorption [23, 27] and the standard spectral shape (mentioned above)  led to the best results in terms of fitting residuals. As seen in Fig. 3 there are fairly good linear correlations (r 2>0.8) when plotting the retrieved parameters (achl(675),adg0) as a function of the observed ones (C, ay0 respectively) obtained either from sample extraction or the actual field measurements.
High correlation is also reached when comparing adg0 to ay0 where no adg0 is available from the in-situ data set. Fig. 4 displays the specific absorption of phytoplankton a*chl which was obtained by dividing the retrieved achl by C obtained from the sample analysis. The variability of achl and its inverse relation to C are consistent with the measurements of Bricaud et al., indicative of packaging and/or pigment compositions effects on a*chl.
5.3 Fitting of reflectance
The fitting of remote sensing reflectance is deemed satisfactory when the fitting residual is small χ2<0.05. To illustrate the diversified ocean conditions encountered in the field measurements, Fig. 5 shows the fitted reflectance (with and without chlorophyll fluorescence) compared to the measured spectra for three extreme scenarios:
1) Extremely high chlorophyll concentration under algae blooming (C over 100mg/m3) leads to the reflectance level below 0.4% over the visible band and the local red peak beyond 700nm; Fig. 5(a)
2) Extremely high CDOM concentration (ay0 is more than 4.5m-1) absorbs so much light in blue-green region that the whole reflectance spectrum is tilted towards longer wavelength; Fig. 5(b)
3) The large NAP fraction of TSS boosts the reflectance level nearly 2% at around 550nm since the highly scattered NAP components dominate in backscattering; Fig. 5(c)
Note, that since in Fig. 5 the retrieved apparent fluorescence contributions are less than 10% of the total reflectance signal these results are excluded from the final quantum yield calculations, as discussed below. Three stations in CB trip with similar values of TSS and ay0 and various C were chosen intentionally for comparison purpose and their fitting results are displayed in Fig. 6. Note that the position of local red peak on reflectance spectra shifted towards longer wavelength (from 685nm in Fig. 6(a) to 695nm in Fig. 6 (c)) with increasing chlorophyll concentration. To illustrate how the quantum yields affect the fluorescence signal on rrs, examples with similar water constituents and two-fold difference in quantum yield are shown in Fig. 7 (η value is 0.41% and 0.79% for LIS01 and LIS13 respectively).
Three typical spectral distributions of the relative fitting residuals are displayed in Fig. 8(a). Generally speaking, the fitted reflectances match well with the measured ones over the whole visible band from 400nm to 750nm. The fitting residuals are small and uniformly distributed over the wavelength space with significant increases at the both ends of the visible spectrum due to low value of the signals at these spectral extremes. More specifically the average relative fitting residuals around 685nm are below 5%. This represents closure of the relation between the inherent optical properties (IOP) and apparent optical properties (AOP) considering that this inversion involves the minimum number of assumptions by employing the comprehensive in-situ hyperspectral data, which are ingested into a rigorous radiative transfer computation. This type of closure has never, to our knowledge, been examined over such diversified coastal water conditions.
The retrieved b̃bp is compared in Fig. 8(b) to the particulate backscattering ratios averaged over 7 channels from the combined ac-s and ECO-bb9 data. Both of them follow the similar trend (r 2=0.66) but a one to one relation is not obtained. Several sources of errors could contribute to this difference. First, up to 10% uncertainties can occur when calibrating the ECO meter in the lab using standard monodisperse spheres . Additionally, the fact that the backscattering coefficient bb is estimated from the scatter at one particular angle (117°) β(117°) assuming proportionality of bb to β(117°) could also introduce up to 10% uncertainties through inter-comparison with the direct multi-angular measurement . Second, the precision of the ac-s data after temperature and scattering correction is reported to be 0.002m-1 for wavelengths below 450nm and 0.01m-1 for wavelengths 450nm and above, however the inaccuracies due to scattering correction for absorption could be as high as 20% . The uncertainties of ac-s not only affect the accuracy of measured b̃bp but also propagate through the radiative transfer program and impact the basic reflectance r̂rs. Finally, although considerable efforts were made to place the fiber probes near the water surface, some small distance between the probe and surface is physically unavoidable, which leads to underestimation of the measured rrs(λ,0-). This error can be significant, especially for highly absorbing and/or scattering water due to abundance of phytoplankton and NAP. Nevertheless the fairly good linear relation of b̃bp between the retrieved values and the measured ones is remarkable and justifies the use of this inversion algorithm, given the many sources of uncertainty discussed above.
5.4 Fluorescence quantum yield
In order to establish confidence in retrieving the fluorescence component from the total reflectance and reduce the errors in η due to reflectance fitting residuals only those stations with retrieved fluorescence magnitude at least 10% of the total reflectance signal at 685nm f·G(685)>0.1·rrs (685) were used for quantum yield calculations. This translates into an upper limit of 50% errors relative to the total retrieved fluorescence amplitude, recalling that our fitting criterion requires χ 2<0.05. There are 29 stations for CB and 21 stations for LIS which met both conditions. In Fig. 9, we present the histogram of the computed quantum yield for CB and LIS. For both campaigns η varies from less than 0.1% to around 1% with nearly 80% of data in the range of 0.2% to 0.5%. This is a relatively small variation considering the large variety of aquatic environments studied here. The mean values of η are 0.32%±0.16% for CB and 0.34%±0.18% for LIS, which is on the same order of those reported in [9, 12] for high irradiance conditions of coastal environments, since most of the measurements were carried out within 2 hours of noon time when the sun is approaching its highest level of the day. In particular defining photosynthetically available radiance (PAR) as: , EPAR ranges from 500 to 1600 µmol.quanta.m-2s-1 with mean value of 1050 µmol.quanta.m-2s-1 for the CB cruise, while in LIS cruise EPAR varies from 900 to 1910 µmol.quanta.m-2s-1 with mean of 1310 µmol.quanta.m-2s-1.
6.1 Validation of approach for quantum yield retrievals using Hydrolight
The approach described here for obtaining the phytoplankton quantum yield from concurrently IOP and AOP measurements requires several assumptions, including a linear response of Hydrolight output to the value of particular backscattering ratio independent of wavelength and depth and homogenous water for relating the fluorescence to Ed, achl, a and bb. In order to test the validity of these assumptions we ultimately fed the retrieved b̃bp, η and achl back into the Hydrolight by including the chlorophyll fluorescence in the run to compare its output of remote sensing reflectance with the measured one as well as the one obtained from the original fitting. Two of our closure results were displayed in Fig. 7 as a dotted line. The resultant reflectance spectra computed from Hydrolight coincide with the fitted ones very well, indicating that our approach is comparable with rigorous radiative transfer program but with less complexity and computation intensity. However, we do notice that there is a slight lower amplitude for Hydrolight output reflectance in the chlorophyll fluorescence region than that of fitted spectra, which can lead to a minor underestimation of the fluorescence quantum yield. This is probably attributable to the fact that a(λem) is used in Eq. (4) for the attenuation of fluorescence radiance instead of diffuse beam attenuation coefficient KLu(λem) for upwelling radiance Lu and in fact a(λem) was approximately 5%–20% smaller than KLu(λem)(λ) depending on the water compositions according to the Hydrolight simulations.
6.2 The natural variability of the quantum yield of phytoplankton
The energy absorbed by the phytoplankton cells dissipates through fluorescence emission, photosynthesis and non-radiative heat transfer process. The phytoplankton quantum yield is related to the photosynthesis rate in a complementary fashion but can be reduced by excessive light stress and nutrient limit since more fractions of active photosynthesis centers become nonfunctional and act as thermally quenching traps [7–9]. Therefore low quantum yields are expected near sea surface for midday conditions. Incorporating all the above quenching factors to the SICF modeling quantum yield is around 1% and its variation is greatly reduced for near noon satellite passing time . Besides, the re-absorption factor Q * tend to be smaller with higher chlorophyll concentration and could reach as small as 0.3 for eutrophic water making the effective quantum yields smaller than 0.5%. This analysis is consistent with our field results presented here. We acknowledge that in a more complete study, sources of error will need more detailed assessment. However, this, we believe, does not detract from the utility of the results presented, which show as a first cut, the type of variations in quantum yield that can result from the diversity of coastal water conditions
6.3 Implication for remote sensing data in productive water
The near surface quantum yield of SICF is of particular interest for remote sensing since the fluorescence signal detected from space and airplane originates from a thin layer of water body given high red light water absorption. On one hand, as seen in Fig. 6, with increasing chlorophyll concentration the red peaks due to elastic scattering are more pronounced and occupy a larger area under the local peaks of the total reflectance with fluorescence included in red region. Under such extreme cases as algal bloom (Fig. 5(a)) the elastic scattering part in red band is responsible for the formation of nearly the whole local maximum with fluorescence signal only filling the left shoulder of the sharp edges of the red peaks. For highly scattered water due to large fraction of NAP in TSS as seen in Fig. 5(c) the fluorescence is only a small fraction in the red bands of reflectance. Under these aquatic environments a large fraction of signals above the baseline is not related to chlorophyll fluorescence but represents a peak in the elastically scattered signal spectrally modulated by a confluence of chlorophyll and water absorption. This leads to over-exaggeration of the quantum yield when FLH is applied to satellite data. It should be also noted however, that SICF satellite data is generally associated with midday clear sky conditions when non-photochemical quenching plays a significant role. Under conditions when FLH is a good estimate of SICF it could be a gross indicator of biomass given the restricted natural variations of its yields. However, our results suggest that the commonly assumed 1% value of SICF quantum yield  might be an overestimation for productive waters.
The sun induced chlorophyll fluorescence (SICF) and its quantum yields in productive water were determined from the in-situ measured surface reflectance spectra as well as absorption and attenuation spectra. The SICF component is obtained from a linear inversion of measured remote sensing reflectance using the spectral shape of fluorescence and elastic scattering. The spectral dependence of elastic scattering part is computed through Hydrolight with the direct field measurements of absorption and attenuation spectra assuming a wavelength independent backscattering ratio. Unlike other fitting techniques this approach removes the commonly assumed power law parameterization for backscattering spectra which is not observed from field data especially for chlorophyll rich water where the scattering spectra are strongly modified by the absorption features of phytoplankton. The quantum yields are calculated using the retrieved parameters including SICF component, phytoplankton absorption and backscattering ratio based on a depth integrated fluorescence modeling verified through Hydrolight. A very diversified coastal waters have been examined with chlorophyll concentration ranging from less than 2 to over 200 mg/m3, ay0 from 0.42 to nearly 5m -1 and total suspended solid (TSS) from 1.7 to 64.8 mg/l. However the derived quantum yields of near surface SICF is highly constrained and varies over a relatively narrow range from less than 0.1% to around 1% with mean values of 0.33%±0.17%.
We would like to express our gratitude to Dr. Alan Weidemann and one anonymous reviewer for the considerable efforts they have put into reviewing this paper and making many helpful suggestions which have helped to greatly improve the revised version. We would also like to thank the following individuals who assisted in field and laboratory measurements in Chesapeake Bay: K. Dillon, B. Elmaanaoui, C. Fan, C. Hladik, R. LaCouture, A. Mustapha, M. Vargas. Laboratory facilities and other logistical support for the Chesapeake Bay campaign, including two research boats, were generously provided by K. Clark at Morgan State University’s Estuarine Research Center at Solomon, MD and by J. Bortz at the Maryland Department of Natural Resources. Finally, we thank Nick Steiner for analyzing the water samples and Ruhul Amin for participating in data collection in Long Island trip 2007. This research was supported partially by grants from NOAA NA06OAR4810162, ONR N00014-08-1-0325 and NOAA NA17AE1624 with subcontract from Environmental Cooperative Science Center, Florida A&M University.
References and links
1. I. G. Fiorani, I. G. Okladnikov, and A. Palucci, “First algorithm for chlorophyll-a retrieved from MODIS-Terra imagery of Sun-induced fluorescence in the Southern Ocean,” Int. J. Remote Sens. 27, 3615–3622 (2006) [CrossRef]
2. F. R. Gower, R. Doerffer, and G. A. Borstad, “Interpretation of the 685nm peak in the water-leaving radiance spectra in terms of fluorescence, absorption and scattering, and its observation by MERIS,” Int. J. Remote Sens. 20, 1771–1786 (1999) [CrossRef]
3. C. M. Hu, F. E. Muller-Karger, C. Taylor, K. L. Carder, C. Kelble, E. Johns, and C. A. Heil, “Red tide detection and tracing using MODIS fluorescence data: A regional example in SW Florida coastal waters,” Remote Sens. Environ. 97, 311–321 (2005) [CrossRef]
4. F. E. Hoge, P E. Lyon, R. N. Swift, J. K. Yungel, M. R. Abbott, R. M. Letelier, and W. E. Esaias, “Validation of Terra-MODIS phytoplankton chlorophyll fluorescence line height. I. Initial airborne lidar results,” Appl. Opt. 42, 2767–2771 (2003) [CrossRef] [PubMed]
5. D. A. Kiefer and W. S. Chamberlin, “Natural fluorescence of chlorophyll a: Relationship to photosynthesis and chlorophyll concentration in the western South Pacific gyre,” Limnol Oceanogr. 34, 868–881 (1989) [CrossRef]
6. Y. Huot, C. A. Brown, and J. J. Cullen, “New algorithms for MODIS sun-induced chlorophyll fluorescence and a comparison with present data products,” Limnol Oceanogr. Methods 3, 108–130 (2005) [CrossRef]
7. M. Babin, A. Morel, and B. Gentili, “Remote sensing of sea surface Sun-induced chlorophyll fluorescence: consequences of natural variations in the optical characteristics of phytoplankton and the quantum yield of chlorophyll a fluorescence,” Int. J. Remote Sens. 17, 2417–2448 (1996) [CrossRef]
8. Y. Huot, C. A. Brown, and J. J. Cullen, “Retrieval of phytoplankton biomass from simultaneous inversion of reflectance, the diffuse attenuation coefficient and Sun-induced fluorescence in coastal water,” J. Geophys. Res. 112, C06013 (2007) [CrossRef]
9. J. R. Morrison, “In situ determination of the quantum yield of phytoplankton chlorophyll a fluorescence: A simple algorithm, observations and a model,” Limnol. Oceanogr. 48, 618–631 (2003) [CrossRef]
10. R. J. Olson, A. M. Chekalyuk, and H. M. Sosik, “Phytoplankton photosynthetic characteristics from fluorescence induction assays of individual cells,” Liminol. Oceanogr. 41, 1253–1263 (1996). [CrossRef]
11. C. S. Roesler and M. J. Perry, “In situ phytoplankton absorption, fluorescence emission, and particulate backscattering spectra determined from reflectance,” J. Geophys. Res. 100, C7, 13279–13294 (1995). [CrossRef]
12. S. Maritorena, A. Morel, and B. Gentili, “Determination of the Fluorescence Quantum Yield by Oceanic Phytoplankton in Their Natural Habitat,” Appl. Opt. 39, 6725–6737 (2000). [CrossRef]
13. A. Gilerson1, J. Zhou, S. Hlaing, I. Ioannou, J. Schalles, B. Gross1, F. Moshary, and S. Ahmed, “Fluorescence component in the reflectance spectra from coastal waters. Dependence on water composition,” Opt. Express 15, 15702–15721 (2007). [CrossRef]
14. A. Gilerson, J. Zhou, M. Ooa, J. Chowdhary, B. M. Gross, F. Moshary, and S. Ahmed, “Retrieval of fluorescence from reflectance spectra of algae in sea water through polarization discrimination: modeling and experiments,” Appl. Opt. 45, 5568–5581 (2006). [CrossRef] [PubMed]
15. A. Gilerson, J. Zhou, and R. Fortich, “Particulate Scattering in Coastal Waters: Chesapeake Bay Study,” Sea Technology 48, 15–18 (2007).
16. Z. P. Lee, K. L. Carder, S. K. Hawes, R. G. Steward, T. G. Peacock, and C. O. Davis, “Model for the interpretation of hyperspectral remote-sensing reflectance,” Appl. Opt. 33, 5721–5732 (1994). [CrossRef] [PubMed]
17. D. Pozdnyakov, A. Lyaskovsky, H. Grassl, and L. Pettersson, “Numerical modeling of transpectral processes in natural waters: implication for remote sensing,” Int. J. Remote Sensing 23, 1581–1607 (2002). [CrossRef]
18. A. Vodacek, S. A. Green, and N. V. Blough, “An experimental model of the solar-stimulated fluorescence of chromophoric dissolved organic matter,” Limnol. Oceanogr. 39, 1–11 (1994). [CrossRef]
19. C. D. Mobley, Light and Water: Radiative Transfer in Natural Waters (Academic Press, 1994).
20. A. Gitelson, J. F. Schalles, and C. M. Hladik, “Remote chlorophyll-a retrieval in turbid, productive estuaries: Chesapeake Bay case study,” Remote Sens. Environ. 109, 464–472 92007).
21. H. R. Gordon, “Diffuse reflectance of the ocean: the theory of its augmentation by chlorophyll a fluorescence,” Appl. Opt. 21, 2489–2492 (1979).
22. J. F. R. Gower, R. Doerffer, and G. A. Borstad, “Interpretation of the 685 peak in water-leaving radiance spectra in terms of fluorescence, absorption and scattering and its observation by MERIS,” Int. J. Remote Sensing , 201771–1786 (1999). [CrossRef]
23. A. Albert and C. D. Mobley, “An analytical model for subsurface irradiance and remote sensing reflectance in deep and shallow case-2 waters,” Opt. Express 11, 2873–2890 (2003). [CrossRef] [PubMed]
24. Z. P. Lee, http://www.ioccg.org/groups/OCAG_data.html.
27. A. M. Ciotti, M. R. Lewis, and J. J. Cullen, “Assessment of the relationships between dominant cell size in natural phytoplankton communities and the spectral shape of the absorption coefficient,” Limnol. Oceanogr. 47, 404–417 (2002). [CrossRef]
28. A. Bricaud, M. Babin, A. Morel, and H. Claustre, “Variability in the chlorophyll-specific absorption coefficients of natural phytoplankton: Analysis and parameterization,” J. Geophys. Res 100, 13321–13332 (1995). [CrossRef]
30. M. Chami, E. Marken, J. J. Stamnes, G. Khomenko, and G. Korotaev, “Variability of the relationship between the particulate backscattering coefficient and the volume scattering function measured at fixed angles,” J. Geophys. Res. 111, C05013 (2006). [CrossRef]
31. J. R. V. Zaneveld, J. C. Kitchen, and C. Moore, “The scattering error correction of reflecting-tube absorption meters,” Proc. SPIE 2258, 44–55 (1994). [CrossRef]