Spectroradiometric Monitoring of Nannochloropsis salina Growth

Algal Research 1 (2012) 22–31

Contents lists available at SciVerse ScienceDirect

Algal Research journal homepage: www.elsevier.com/locate/algal

Spectroradiometric Monitoring of Nannochloropsis salina Growth Thomas A. Reichardt a,⁎, Aaron M. Collins b, Omar F. Garcia b, Anne M. Ruffing b, Howland D.T. Jones b, Jerilyn A. Timlin b a b

Remote Sensing and Energetic Materials Department, Sandia National Laboratories, P. O. Box 969, MS 9056, Livermore, CA 94551 Bioenergy and Defense Technology Department, Sandia National Laboratories, P. O. Box 5800, MS 0895, Albuquerque, NM 87185

a r t i c l e

i n f o

Article history: Received 21 September 2011 Accepted 9 December 2011 Available online 11 February 2012 Keywords: algal biofuel algal growth remote sensing hyperspectral reflectance

a b s t r a c t The high productivity of fluidically mixed open ponds for algal biofuel production is accompanied by high environmental and temporal variability. Therefore, a recognized need exists for rapid monitoring of open ponds to quantify algal growth rates, assess algal stress, detect the presence of invading species, and determine the optimum time for harvesting. Multispectral/hyperspectral approaches are now being used to optimize conventional agriculture practices and similar remote sensing techniques can potentially be used for monitoring algal ponds. In this work, we assess the application of remote techniques for algal biofuel production by using a dual-channel spectroradiometer to monitor the laboratory-scale growth of Nannochloropsis salina, a popular microalgal candidate for biofuels. One channel of the spectroradiometer measures the downwelling irradiance while the second channel simultaneously monitors the upwelling radiance, and the reflectance is calculated by ratioing these two signals. A detailed reflectance model is developed to interpret the acquired spectra, enabling a remote assessment the culture's optical depth as well as the relative optical activity of different algal pigments in N. salina. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The fluidically mixed open ponds currently being considered for algal biofuel production are typically ~15-cm deep and cover 1000–5000 m 2 [1]. Circulation and mixing of a raceway pond using paddlewheels can yield a 10-fold improvement in productivity compared to unmixed ponds [2]. This high productivity is, however, accompanied by high temporal variability: while decisions concerning fertilizing, harvesting, or weed control in conventional agriculture are made on a time scale of days, similar decisions concerning mixed open ponds must take place within hours, and mistakes can result in total culture loss [3]. Open ponds are inherently susceptible to temperature variations and invasions by other algae, algae grazers, and bacteria/fungi/amoeba [2]. Furthermore, the paddlewheels do not maintain turbulent mixing [4], allowing for potential spatial variation of growth conditions over the pond volume. Thus, a recognized need exists for rapid, spatially resolved monitoring of pond characteristics to quantify algal growth rates, detect algal stress and the presence of an invading species, and identify the optimum time for harvesting. Such monitoring may be achieved by remote sensing techniques, of which perhaps the most promising are the multispectral/hyperspectral remote spectroradiometric approaches now being used to optimize conventional agricultural practices [5–8].

⁎ Corresponding author. Tel.: + 1 925 294 4776; fax: + 1 925 294 2595. E-mail address: [email protected] (T.A. Reichardt). 2211-9264/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.algal.2011.12.001

In this work, we assess the value of monitoring algal cultures with remote spectroradiometry by measuring the growth cycle of Nannochloropsis salina, a prime candidate algal species for biofuel production. This laboratory-scale experiment measures the spectrally resolved reflectance of a culture, and a detailed reflectance model is developed to interpret the acquired spectra. Analyzing the temporal variation of the spectrally resolved reflectance, we demonstrate the ability to remotely determine the optical depth, anticipated as a surrogate measurement of the dry weight, as well as the relative optical activity of carotenoids compared to that of Chlorophyll a (Chl a), providing a signature relevant to algal health. In our assessment of spectroradiometric monitoring for algal biofuel production, we rely heavily on the body of work spanning several decades to develop remote sensing approaches for natural bodies of water (e.g., for climate change studies [9]). Over this time, significant effort has been invested towards inverting reflectance models to interpret the inherent optical properties (see [10] and references therein) of water bodies, specifically the backscattering coefficient bb (m -1) and the total absorption coefficient a (m -1), from upwelling radiance measurements. Achieving this goal continues to challenge the community: it has been noted that inversion of reflectance models for water bodies can be an “ill posed problem” [11], meaning that multiple forward solutions can provide similar reflectance spectra. Acknowledging the uncertainties associated with inverting reflectance models [12], we also anticipate that the model parameters will be much more constrained for engineered ponds than they are for natural bodies of water. Several of the unknowns encountered when

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

analyzing data for oceanography (e.g., the water depth, the bottom reflectivity, and the algal species present) can be measured or estimated with reasonably high fidelity for engineered open ponds, and constraining the model will significantly reduce the ill-posed nature of the model inversion process [13]. The specific algae we target for this study is N. salina. The members of the genus Nannochloropsis have garnered interest as biofuelcandidate microalgae because of their high lipid content (20-50% dry weight) and lipid productivity [14]. Noteworthy to both in-situ and remote spectral assessment, Nannochloropsis sp. have been observed to exhibit significant changes in the optical activity of carotenoids with respect to Chl a (Nannochloropsis sp. contain neither Chl b nor Chl c) in response to nitrogen limitation [15–19]. This effect is not unique to Nannochloropsis sp., and has been observed for other alga as well [20–24]. The ability to remotely assess such changes in optical properties could provide a robust monitoring tool. Remote sensing of Nannochloropsis sp. has been pursued in a prior effort very relevant to our work: Gitelson et al. [3] evaluated the reflectance of Nannochloropsis sp. and derived simple expressions relating the dry weight and chlorophyll content per algal cell based on the relative strengths of two spectral bands, one red (670–680 nm) and the other near-infrared (750–950 nm). In that work, Gitelson et al. [3] decomposed the Nannochloropsis sp. absorption spectrum into a series of Gaussian features, a method we employ here as well. We expand upon this prior work to demonstrate a spectral decomposition algorithm for application to spectroradiometric monitoring of algal biofuel production, combining the procedures of Gitelson et al. [3] with a reflectance model based on the work of Lee et al. [25]. To this reflectance model, which includes the effects of a limited water column and the bottom reflection, we add the contributions of specular reflections and Chl-a fluorescence. We then apply the model to the reflectance spectra acquired in the laboratory-scale experiment to solve for the temporal variation of the inherent optical properties of the algal culture. Results from the model are compared to offline measurements made on extracted samples. While the reflectance model presented in this paper is specific for N. salina, it could be readily modified for application to other alga with other pigments as well, providing rapid and specific information of algal cultures with largescale mixed pond applicability.

23

Fig. 1. Spectrally resolved irradiance of laboratory fluorescent lighting.

looking fiber allows for the full collection of downwelling light. Downwelling and upwelling spectra were recorded every five minutes. Typically both channels are set to a 10-ms integration time, and 100 spectra are averaged for each acquisition. The relative spectral responses of both channels were determined with an Ocean Optics LS-1 tungsten-halogen calibrated light source. After completing the measurements, the culture is removed from the beaker and replaced with a 51-mm diameter Spectralon calibrated reflectance target (LabSphere SRS-20-020, North Sutton, NH, USA) positioned at the prior height of the water surface. Acquiring reflectance data on this reference target allows us to calibrate the measurement approach and report absolute reflectance values. This paper reports on reflectance measurements acquired over a period of 11 days, beginning two hours after the inoculation and lasting through the exponential growth phase and well into the stationary phase. In parallel to measuring the reflectance, we also sampled the culture for offline analysis, on Days 0–4 and 7–10, typically extracting samples 90 minutes after the lighting was cycled on. For each of these samples, we recorded the pH, acquired photosynthesis-yield readings with a Walz (Effeltrich, Germany) mini-PAM, and measured the OD750 with the Lambda Bio spectrophotometer. For four of these nine samples (on Days 2, 4, 7, and 9), we also measured the spectrally resolved (300–900 nm) OD with a Beckman Coulter (Brea, CA, USA) DU 800 Spectrophotometer.

2. Materials & Methods 2.2. Reflectance model for spectral decomposition 2.1. Experimental apparatus The N. salina innoculum is grown at room temperature with shaking (150 rpm) in a 500-mL baffled flask containing 100 mL of F/2 media. When the innoculum reached a 750-nm optical depth (OD750) of ~ 1.0, as measured with a Perkin Elmer (Waltham, MA, USA) Lambda Bio spectrophotometer, the culture is transferred to 300 mL of fresh F/2 media in a cellophane-covered beaker to obtain a starting OD750 of 0.1. The culture is mixed using a magnetic stir bar, and a mixture of air and 1% CO2 is bubbled through a pipette attached to the side of the beaker. Fluorescent lighting provides a spectrally structured irradiance (see Fig. 1) of 87 μmol photons/s-m 2 (18 W/m 2), operated with a 12-hours-on / 12-hours-off diurnal cycle. A piece of black cardboard is positioned under the beaker to reduce reflections from the bottom surface. The downwelling and upwelling light are measured with an Ocean Optics (Dunedin, FL, USA) Jaz dual-channel fiber-coupled spectroradiometer. To capture the upwelling light, one channel of the spectroradiometer is connected to a bare 600-μm core-diameter fiber directed downward over the beaker. The other channel is connected to a 50-μm core-diameter fiber with a cosine-corrector diffuser attachment directed upward to measure the downwelling light. The 0.22-numerical aperture (NA) of the bare fiber defines a 2 × sin -1(0.22) = 25 o field-of-view for the downwardlooking fiber, while the cosine corrector attached to the upward-

Our reflectance model applies the analytical equations presented by Lee et al. [25] to relate the reflectance of the algal culture to the optical activity of the specific N. salina pigments. We provide here all of the expressions necessary to perform our spectral decomposition, but we refer the reader to the paper by Lee et al. [25] and references therein for details concerning the derivations of these equations. 2.2.1. Spectral inputs for the model Following a method originally presented by Gorden et al. [26] and since widely used in water reflectance models [12,25,27–48], we parameterize rbs, defined as the ratio of the water-leaving radiance just below the water surface Lw (W/m 2-nm-sr) to the downwelling irradiance just below the water surface Ed- (W/m 2-nm), as a function of u, r bs ¼

−

Lw ¼ f ðuÞ: Ed −

ð1Þ

The parameter u is defined by the backscattering coefficient bb and the absorption coefficient a as u¼

bb ; a þ bb

ð2Þ

24

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

where both bb and a are wavelength-dependent. The term rbs is the below-surface radiance reflectance with units of sr -1. In Section 2.2.3, we will calculate the irradiance reflectance, which is the unitless ratio of the upwelling irradiance to the downwelling irradiance. To assist the reader in differentiating between these two reflectance values, we designate radiance reflectance (sr -1) with the letter “r,” while using the Greek letter “ρ” for the unitless irradiance reflectance. These terms, and all other symbols in this paper, are defined in Table 1. Seeking to express a in Eq. (2) as the sum of optically active pigment contributions, the absorbance spectrum of N. salina is assumed identical to the Nannochloropsis sp. absorbance spectrum presented in Fig. 3 of Gitelson et al. [3]. Again following Gitelson et al. [3], we approximate this absorbance spectrum as the sum of Gaussian features (see [49–52] for other examples of this method), with a Gaussian

Table 1 Variable units and definitions. Variable Acarotenoids AChl a A525 nm A557 nm a awater B bb C1 C2 C3 C4 C5 C6 Dd DuB DuC Ed+

Units

m-1 m-1 m-1 m-1 m-1 m-1 m-1 m-1 W/m2-nm

W/m2-nm

Ed-

W/m2-nm

F Fot

W/m2-nm

H j Lf

m radians m

Lw

Q

W/m2-nmsr sr

ras

sr-1

e ras

sr-1

rbs B rbs

sr-1 sr-1

C rbs

sr-1

od rbs

sr-1

u κ λ λ0 ρas ρB ρsr

m-1 nm nm

Definition Normalizeda absorbance of carotenoids Normalizeda absorbance of Chl a Normalizeda absorbance by 525-nm feature Normalizeda absorbance by 557-nm feature Absorption coefficient Pure water absorption coefficient Unity-normalized average backscatter by phytoplankton Backscattering coefficient Scaling factor for Chl-a absorption Scaling factor for carotenoid absorption Scaling factor for absorption by 525-nm feature Scaling factor for absorption by 557-nm feature Scaling factor for backscattering coefficient Scaling factor for fluorescence Downwelling distribution Upwelling distribution from bottom reflection Upwelling distribution from water column scatter Downwelling irradiance immediately above the water surface Downwelling irradiance immediately below the water surface Fluorescence collected by downward-looking detector Unity-normalized fluorescence spectrum for optically thin conditions Water depth Below-surface solar zenith angle Effective pathlength through which fluorescence propagates Upwelling radiance immediately below the water surface Ratio of irradiance to radiance for light emitted from water surface Above-surface radiance reflectance resulting from both volumetric scattering and the bottom reflection Above-surface radiance reflectance from elastically scattered light Total below-surface radiance reflectance Below-surface radiance reflectance from the bottom surface Below-surface radiance reflectance resulting from volumetric scattering from the water column Below-surface radiance reflectance for optically deep water Ratio of backscattering coefficient to sum of backscattering and absorption coefficients: bb / (bb + a) Quasi-diffuse attenuation coefficient: bb + a Wavelength of light Center wavelength of Gaussian feature Total above-surface irradiance reflectance Irradiance reflectance of bottom surface Irradiance reflectance for specular reflections

a The absorbance spectra for the four spectral components are normalized so that the nominal sum of the spectra is peak-normalized to unity [see Fig. 2(a)].

feature of center wavelength λ0 (nm) and standard deviation σ (nm) expressed as
   1 λ−λ0 2 : feature ðλÞ ¼ amplitude  exp − 2 σ

ð3Þ

We obtain a good match with Gitelson et al.'s [3] absorbance spectrum by including eleven such Gaussian features [see Fig. 2(a)]. Of these eleven features, seven of them (centered at 422, 438, 584, 628, 668, 682, and 697 nm) are attributed to Chl a, while two features (centered at 463 and 493 nm) are attributed to carotenoids. In addition to absorbance features associated with Chl a and carotenoids, Nannochloropsis sp. exhibit absorbance between 520–560 nm as well [3,53], which we account for by including two additional features centered at 525 and 557 nm. Fig. 2(b) displays our assumed grouping of the features into “Chl a” and “carotenoid” along with the two additional features, while Table 2 lists the parameters associated with the Gaussian features. The value of a in Eq. (2) can now be expressed as the sum of five terms, a ¼ C 1 AChl a þ C 2 Acaroteoids þ C 3 A525 nm þ C 4 A557 nm þ awater ;

ð4Þ

where C1-C4 (m -1) are wavelength-independent scaling factors and
 
    1 λ−422 nm 2 1 λ−438nm 2 AChl a ¼ 0:801⋅ exp − þ 0:214⋅ exp − 2 26 nm 2 8 nm  
 
  1 λ−584 nm 2 1 λ−628 nm 2 þ 0:185⋅ exp − þ0:081⋅ exp − 2 16 nm 2 18 nm
 
    1 λ−668 nm 2 1 λ−682 nm 2 þ0:408⋅ exp − þ 0:511⋅ exp − 2 10 nm 2 9 nm
   1 λ−697 nm 2 þ0:031⋅ exp − ; ð5Þ 2 15 nm

Fig. 2. Absorbance features incorporated into the reflectance model. The absorbance spectrum for N. salina is developed by approximating the Nannochloropsis sp. absorbance spectrum presented in [3] to a sum of Gaussian components (a), which are then divided into four different spectral features (b).

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31


   1 λ−463 nm 2 Acarotenoids ¼ 0:335⋅ exp − 2 17nm
   1 λ−493 nm 2 ; þ0:446⋅ exp − 2 15 nm

ð6Þ


   1 λ−525nm 2 ; A525 nm ¼ 0:147⋅ exp − 2 18nm

ð7Þ

and
   1 λ−557 nm 2 : A557 nm ¼ 0:018⋅ exp − 2 16 nm

ð8Þ

The water absorption spectrum awater is taken from Smith and Baker [54]. When applying reflectance models to natural bodies of water, it is often assumed that the spectral variation of bb can be adequately represented by a simple power-law dependence (i.e., bb ∝ λ -n) [22,55,56]. Such a relationship, describing a scattering cross section that increases with decreasing wavelength, is reasonable for natural water bodies because the majority of backscatter originates from suspended solids (e.g., detritus) that do not exhibit distinct spectral features in the visible wavelength range. However, the primary source of elastic backscatter in algal ponds will be the algal cells, and this backscatter can possess distinct spectral features due to variations in the algal refractive index [49,57–63]. Without the specific backscattering spectrum for N. salina, we use the generic phytoplankton backscattering spectrum that Lahet et al. [61] calculated by averaging the nine backscattering spectra measured by Ahn et al. [60]. We extrapolate  the and normalize this spectrum to unity at 750 nm to generate B, normalized average backscattering spectrum for phytoplankton. The value of bb is calculated by multiplying this normalized spectrum by scaling factor C5 (m -1),  bb ¼ C 5 B:

ð9Þ

2.2.2. Subsurface elastically scattered light 2.2.2.1. Scattering from the water column. Having defined wavelengthdependent expressions for the inputs to u, rbs can now be parameterized as a function of u. To account for turbid (large u) waters, Lee et al. [25] used the code Hydrolight [64,65] to numerically solve the radiative transfer equation [66] for bodies of water, finding that for optically deep water (i.e., in which a negligible fraction of light reaches the bottom),   od 0:752 r bs ≈ 0:070 þ 0:155u u:

ð10Þ

Table 2 Decomposition of the N. salina absorbance spectrum.

While Eq. (10) is applicable to optically deep water, the assumption of optically deep water is easily violated in algal raceways, and especially in our laboratory measurement. Compared to optically deep water, the remote sensing reflectance for optically shallow water will be decreased by the reduction in scatter due to the limited water-column depth. On the other hand, the reflectance will now include a contribution from the bottom reflection. We account for both of these effects using the relationships derived by Lee et al. [25,28]. First, accounting for the reduced water depth, Lee et al. [25,28] C expressed the reflectance due to the water column, rbs as C

r bs ≈rbs

od

n h   io C 1−1:03 exp − Dd þ Du κH ;

Center wavelength of Gaussian λ0 (nm)

Standard deviation of Gaussian σ (nm)

Amplitude

Pigment

1 2 3 4 5 6 7 8 9 10 11

422 438 463 493 525 557 584 628 668 682 697

26 8 17 15 18 16 16 18 10 9 15

0.801 0.214 0.335 0.446 0.147 0.018 0.081 0.185 0.408 0.511 0.031

Chl a Chl a Carotenoids Carotenoids

Chl Chl Chl Chl Chl

a a a a a

ð11Þ

where κ, the quasi-diffuse attenuation coefficient (m -1), is the sum of the backscattering and absorption coefficients, κ ¼ a þ bb ;

ð12Þ

while H is the water depth (m). In Eq. (11), Dd is the vertically averaged downwelling distribution approximated as Dd ≈1= cosðjÞ;

ð13Þ

where j is the below-surface solar zenith angle (rad). The vertically averaged distribution DuC of the upwelling radiance from the water-column scattering will be distributed over a broader range of angles than the downwelling light. By matching the results of Hydrolight simulations to an expression based on the work of Kirk [67], Lee et al. [25] found that DuC could be well described by C

0:5

Du ≈1:2  ð1 þ 2:0uÞ

:

ð14Þ

2.2.2.2. Bottom reflection. Considering now the interaction of the light with the bottom of the water column, Lee et al. [25,28] treat the bottom as a Lambertian reflector. In our measurement of the laboratory algal culture, we detect bottom-surface reflections from both the glass beaker and the black cardboard underneath. While the black cardboard is expected to be a somewhat diffuse reflector, the glass bottom of the beaker generates a significant specular reflection. We nevertheless assume that the bottom can be treated as a Lambertian reflector with irradiance reflectance ρ B, so the reflectance contribution from the bottom can be expressed as [25] h   i B B B r bs ≈0:31ρ exp − Dd þ Du κH ;

ð15Þ

Eq. (15) accounts for the exp(− DdκH) attenuation of the downwelling light and the exp(− DuBκH) attenuation of the bottomreflected upwelling light. Matching again to Hydrolight results, Lee et al. [25] provide the approximation B

Feature number

25

0:5

Du ≈1:1  ð1 þ 4:9uÞ

:

ð16Þ

The total below-surface reflectance is the sum of the contributions from the water column and bottom reflection, C

B

r bs ¼ r bs þ r bs :

ð17Þ

2.2.3. Calculating the above-surface irradiance reflectance Eq. (17) provides a relationship for the below-surface reflectance, but the reflectance measured by a remote sensing instrument is instead the above-surface ratio of the water-leaving radiance to the downwelling irradiance. For nadir viewing, the above-surface reflectance resulting

26

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

e from below-surface elastically scattered light ras (sr-1) can be approximated as [25]

e ≈ r as

0:518r bs : 1−1:562r bs

ð18Þ

As discussed in Section 2.1, we calibrate our reflectance measurements against a reference target, the wavelength-resolved reflectance e is a of which is reported as a unitless irradiance reflectance, while ras -1 radiance reflectance with units of sr . We therefore convert to the e e irradiance reflectance ρas , multiplying ras by the upwelling radiance-toirradiance conversion factor Q, e e ρas ¼ Q ⋅r as :

ð19Þ

the approach that Ahmed et al. [75] used to account for fluorescence absorption, and incorporating the attenuation by scattering as well, we include the collected fluorescence component F (W/m 2-nm) as   ot F ¼ C 6 F exp −κLf ;

where F ot is a unity-normalized Chl-a fluorescence spectrum acquired under optically thin conditions (see Fig. 3), Lf is the effective length (m) over which fluorescence is scattered and absorbed, and C5 is a scaling constant independent of wavelength (W/m 2-nm). The final expression for the total above-surface irradiance reflectot tance ρas sums the contributions from (1) the below-surface elastically scattered light from both the water column and bottom reflection, (2) the specular reflectance from both the water surface and cellophane cover, and (3) the Chl-a fluorescence, resulting in sr

Values of Q are observed to span between 3.2 and 5.1 rad [68]: we again follow Lee et al. [25] who set Q = 3.25 rad. 2.2.4. Water-surface reflection and Chlorophyll fluorescence The above equations consider only the elastically scatted light originating from below the water surface. Two additional contributions to our collected light must be considered: (1) the specular reflections from the water surface and cellophane cover and (2) the fluorescence originating from Chl a. Considering first the former contribution, the collection of specular reflections from the water surface can constitute a significant fraction of the total upwelling radiance [69]. The contribution by specular reflections can be reduced by orientating the downward-looking sensor at an angle [70–72], but this is not practical for our experimental configuration. So, identical to Gould et al. [73], we take advantage of the fact that the refractive indices for both water and cellophane are relatively constant from 450–720 nm and treat the contribution from specular reflections as a spectrally flat irradiance reflectance ρ sr. Depending on the spectral structure of the incident irradiance, Chl-a fluorescence can also contribute significantly to the reflectance spectrum. Chl-a fluorescence is particularly noticeable in our reflectance spectra because the fluorescence (see Fig. 3) matches a spectral region of limited incident irradiance (see Fig. 1). At the wavelengths corresponding to the lowest levels of incident irradiance (~690 nm and ~ 700 nm), the Chl-a fluorescence can well exceed the upwelling elastically scattered light. Modeling of this fluorescence feature is complicated by the effect of self-absorption. Because Chl a absorbs in the same spectral region that it fluoresces, the detected fluorescence from Chl a can be distorted in relation to the spectrum that would be obtained under optically thin conditions [74]. Following

e ρas ¼ ρas þρ þ

F : Ed þ

ð21Þ

2.2.5. Applying the reflectivity model to the laboratory measurement geometry As discussed in Section 1, determining inherent optical properties by inversion of reflectance models is challenging because multiple forward solutions can result in similar reflectance spectra. For our laboratory beaker, and for future monitoring of engineered open ponds, we can improve the rigor of an inverse solution by reducing the number of parameters that are allowed to vary. For our laboratory beaker measurements, a number of parameters can be set and held constant with reasonable accuracy (see Table 3). With the fluorescent lighting positioned directly above the laboratory beaker, we can approximate the light as originating from the zenith direction, and therefore j = 0, so Dd = 1. At the beginning of the experiment, the water depth was 2.8 cm, but due to evaporation and sampling loss the depth decreased ~0.4 mm/day. We set H to the initial value of 2.8 cm. The irradiance reflectance of the bottom surface is set to the measured value of ρ B = 0.30. We expect a minimum value of ρ sr = 0.02 from the water surface [69], and we assume this is valid for Day 0. However, after extracting a sample on Day 1, the downward-looking sensor detected a ~0.01 increase in ρ sr as evidenced by an abrupt, wavelength-independent increase in the reflectance, likely due to partial collection of the specular reflection from the water surface or cellophane cover. A similar effect was noted after sampling on Day 10, with another corresponding ~0.01 increase in ρ sr. We therefore set ρ sr = 0.02 for Day 0, ρ sr = 0.03 for Days 1–9, and ρ sr = 0.04 for Day 10, with the value of ρ sr specifically transitioning at the times corresponding to sample extraction on Days 1 and 10. The complication introduced by potential specular reflections from the cellophane cover is specific to the laboratory-scale system and will not be an issue for monitoring outdoor cultures. When applying Eq. (21) to invert an acquired reflectance spectrum, we allow C1-C6 and Lf to vary, resulting in a total of seven wavelength-independent parameters that are adjusted to optimize the quality of the fit. This optimization is performed via MATLAB's fminsearch function, which implements the simplex search method

Table 3 Values set constant when inverting reflectance model.

Fig. 3. Chl-a fluorescence spectrum from N. salina acquired with a home-built fluorescence instrument. The spectrum is referenced to an absolute calibration source and is displayed with a vertical scaling linearly proportional to units of W/m2-nm-sr.

ð20Þ

Variable Value

Justification

j H ρB ρsr

Zenith light source Water height at time of innoculation Measured for a beaker filled with water Value consistent with the minimum water surface reflection, with additional contribution from specular reflections included after sampling on Days 1 and 10.

0 rad 2.8 cm 0.30 0.02, 0.03, or 0.04

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

of Lagarias et al. [76] to minimize the normalized error between the reflectance spectrum and the model over the 450–720 nm spectral range.

3. Results and Discussion 3.1. Temporal variation of model parameters Applying this inversion model to our reflectance measurements, we obtain reasonable agreement with the spectral features over the entire time span of the experiment. Fig. 4 displays the reflectance spectra acquired at approximately 90 minutes after the lights are cycled on each morning, with each measured spectrum accompanied by the modeling results and the fit residual displayed below. Fig. 5 displays the temporal variation of C1-C6 and Lf over the 11 days of the experiment.

27

Three main observations are evident from the temporal histories of the model parameters: (1) The scaling factors C1 and C2, associated with optical activity of Chl a and carotenoids, respectively, exhibit an initial exponential rise from Day 0 to Day 4, after which C1 gradually decays while C2 remains relatively constant. (2) The scaling factor C5, the only parameter associated with the backscattering coefficient, exhibits an initial exponential rise from Day 0 to Day 4, followed by a more gradual rise. (3) The fluorescence scaling factor C6 initially increases while the characteristic fluorescence length Lf initially decreases, after which both values remain relatively constant over the remaining time of the experiment. The trend observed for C1 and C2 is consistent with an exponential growth phase that lasts five days, after which the culture transitions to stationary phase. During the exponential growth phase, the optical

Fig. 4. Irradiance reflectance spectra (gray) compared to model fits (black). The residual (data-model) is displayed with a dotted line below each reflectance spectrum. These spectra correspond to ten sequential days, Days 1–10 labeled (a) to (j) respectively, based on the data acquired ~ 90 minutes after the lights were cycled on each day.

28

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

Fig. 5. Temporal dependence of the calculated model parameters. Breaks in the data correspond to the night photoperiod when the laboratory lighting is turned off. See text for further explanation of parameter details.

activity of pigments should follow the expected increase in cell concentration. Because Chl a plays a central role in photosynthesis, we also expect the carotenoid-to-Chl-a ratio to reach a minimum value during the exponential phase and to then increase upon entering stationary phase [see Fig. 6(a)]. Fig. 6(b) displays a comparison between the value of C1 and an estimate of the Chl-a absorption based on the spectrally-resolved OD measurements performed with the Beckman Coulter DU 800 Spectrophotometer. Here the Ch-a absorption, as measured with the DU 800, is assumed proportional to OD680 – OD750, the OD difference between 680 nm (at the peak of Chl-a absorption) and 750 nm (corresponding to negligible absorption by N. salina). The linear scaling of the two vertical axes has been set to illustrate the agreement between the reflectance-based results and the spectrophotometer measurements. Similar to pigment absorption, the backscatter scaling factor C5 is expected to increase during the exponential growth phase. Fig. 6(c) displays a comparison between the value of C5 and OD750 measurements by both spectrophotometers. As with Fig. 6(b), the relative linear scalings of the vertical axes have been set to illustrate agreement between the multiple measurements. The agreement between the reflectance-based backscatter coefficient and the OD750 measurements demonstrates the ability to remotely assess the optical depth from reflectance measurements. Considering the Chl-a fluorescence, the magnitude of fluorescence for an optically thin sample is expected to increase linearly with increasing algal concentration, because more algal cells are available to absorb and emit light. As optically thick conditions are reached late in the growth cycle, the fluorescence is expected to be relatively independent of algal concentration because any increase in algal cells will decrease the penetration depth of light into the sample. In other words, the fluorescence increase with concentration will be effectively offset by the decreased optical penetration depth into the sample. This phenomenon is illustrated by the spectra in Fig. 4 and the Chl-a

absorption (C1), cell backscatter (C5), and Chl-a fluorescence (C6) scaling factors in Fig. 5. Beyond Day 5, the backscatter, which serves as an indicator of cell concentration, increases over time, suggesting the culture is in the mid- to late-exponential phase. However, the Chl-a absorption and fluorescence scaling factors for the same time period level off and even decrease for Chl-a absorption, indicating that these parameters no longer scale linearly with the increasing cell concentration. 3.2. Limitations of the model inversion method Despite our success at applying the reflectance model, we recognize that the model inversion process likely limits the fidelity of the calculated model parameters. The relationship between parameter u and the inherent optical properties a and bb is unfortunately not a bijection [11], but rather proportional changes in a and bb will effectively offset each other in the calculation of u [see Eq. (2)]. Our reflectance model considers algal cells as the only volumetric backscattering component in the culture, and the algal cells function as the primary absorber over the 450–720 nm spectral range of the downwelling irradiance (see Fig. 1). With both a and bb scaling similarly with increasing algal concentration [77,78], both the numerator and denominator in Eq. (2) scale similarly as well, effectively ratioing out the dependence of u upon the cell concentration. As a result, the main effect of increasing the cell concentration is to decrease the relative contribution of the bottom reflection (which dominates in the early-growth, optically thin regime) in relation to the scatter from the water column (which contributes much more to the reflectance in the late-growth, optically thick regime). To graphically display this point, Fig. 7(a) and (b) present intermediate calculations used to generate the reflectance spectra of Fig. 4(a) and (j), respectively. Fig. 7(a) shows the reflectance contribution from the water column scattering is negligible early in the growth cycle, while by the end

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

29

Fig. 7. Radiance reflectance contributions of the bottom reflection and water column both (a) early and (b) late in the growth cycle. The results in (a) correspond to the reflectance spectrum in Fig. 4(a) while the results in (b) correspond to the spectrum in Fig. 4(j).

sides of the beaker, so to first order the one-dimensional assumption holds. However, as discussed in Section 2.2.2.2, the reflection from the bottom of the beaker is expected to have a significant specular component, likely impacting the effective pathlength through which the reflected light propagates, and compromising our use of Eq. (15).

3.3. Future efforts Fig. 6. Analysis of model parameters: (a) displays the ratio of carotenoid-to-Chl-a optical activity, and (b) and (c) display spectrophotometer measurements compared to Chl-a absorption and algal backscatter, respectively.

of the growth cycle [Fig. 7(b)] the contribution by the water column scattering is comparable to that of the bottom reflection. In addition to ambiguities arising from the inversion process, there are simplifying approximations made in the reflectance model deserving further analysis. The model currently relies solely on the spectra of Gitelson et al. [3] to represent the N. salina absorbance spectrum needed to calculate a in Eq. (2). We allow for variation of this absorbance spectrum in a somewhat ad-hoc manner, letting the four different spectral components vary only in relative magnitude. Thus effects such as spectral shifting and broadening, which may be significant, are not currently included. Laboratory measurements of the spectrally resolved absorption with a well-characterized spectrophotometer would help assess the need to include such effects. Furthermore, when compared to the approximations in our calculation of a, our treatment of bb is even less rigorous. Our approximation of the backscatter as an average spectrum determined from measurements of multiple phytoplankton [see Eq. (9)] resulted from a lack of any detailed information on the specific spectral variation of backscattering from N. salina. In the future, we could consider directly measuring the backscatter coefficient using existing techniques (see [79] and references therein). Finally, the derivations of Lee et al. [25] assume a one-dimensional water column terminated in a Lambertian reflector. The FOV of the downward-looking detector did not extend to the

Despite the limitations mentioned above, the results presented herein highlight the utility and potential of using spectrally resolved reflectance in open systems. By combining reflectance measurements with a reflectance model that accounts for the spectrally varying absorption and scatter, we have demonstrated that spectral signatures exist to remotely monitor the optical depth of an algal culture as well as the relative optical activity of algal pigments. Because the inherent optical properties can be derived from a reflectance spectrum in only a few seconds, reflectance measurements could provide a realtime monitor of chlorophyll concentration and algal growth rate in the laboratory. The method we have presented is both simple to set up in the laboratory as well as cost-competitive, with significant advantages over flow-cell based real-time culture analysis methods that often foul with algae during even moderate use. In addition to highlighting the utility in the laboratory, the results of this study motivate transitioning this work to monitor a larger-scale outdoor culture system. This transition will be accompanied by new challenges such as the variable angle of the sun and the different downwelling irradiances from the sun and the diffuse sky [80] – effects that will be incorporated into the reflectance model. Also, while this work focused on the identification of signatures using a single-FOV system, in different configurations hypserspectral remote sensing can achieve broad area, spatially resolved coverage from a fixed or mobile platform [5–8]. The reflectance model demonstrated in this work can be applied to the pixel-resolved data from a visible hyperspectral imager, enabling spatially resolved measurements of inherent optical properties in mixed algal ponds. This and other options will be

30

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31

considered when addressing how to best meet research as well as commercial needs.

Acknowledgments We gratefully acknowledge the technical assistance of Lindsey Gloe, Christine Trahan, and Kylea Parchert (all of Sandia National Laboratories, NM, USA) in performing the sampling measurements. This work was supported by Laboratory Directed Research and Development program at Sandia National Laboratories. Sandia is a multi-program laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.

References [1] A. Richmond, S. Boussiba, A. Vonshak, R. Kopel, A new tubular reactor for mass production of microalgae outdoors, Journal of Applied Phycology 5 (1993) 327–332. [2] J.R. Benemann, Opportunities and challenges in algae biofuels production, A position paper in line with Algaeworld 2008, Sept. 2008. [3] A.A. Gitelson, Y.A. Grits, D. Etzion, Z. Ning, A. Richmond, Optical properties of Nannochloropsis sp and their application to remote estimation of cell mass, Biotechnology and Bioengineering 69 (2000) 516–525. [4] A. Gitelson, H. Qiuang, A. Richmond, Photic volume in photobioreactors supporting ultrahigh population densities of the photoautotroph Spirulina platensis, Applied and Environment Microbiology 62 (1996) 1570–1573. [5] S. Moran, G. Fitzgerald, A. Rango, C. Walthall, E. Barnes, W. Bausch, T. Clarke, C. Daughtry, J. Everitt, D. Escobar, J. Hatfield, K. Havstad, T. Jackson, N. Kitchen, W. Kustas, M. McGuire, P. Pinter Jr., K. Sudduth, J. Schepers, T. Schmugge, P. Starks, D. Upchurch, Sensor development and radiometric correction for agricultural applications, Photogrammetric Engineering and Remote Sensing 69 (2003) 705–718. [6] J.L. Hatfield, A.A. Gitelson, J.S. Schepers, C.L. Walthall, Application of spectral remote sensing for agronomic decisions, Celebrate the Centennial [Supplement to Agronomy Journal] (2008) S-117–S-131. [7] J.A.J. Berni, P.J. Zarco-Tejada, L. Suárez, E. Fereres, Thermal and narrowband multispectral remote sensing for vegetation monitoring from an unmanned aerial vehicle, IEEE Transactions on Geoscience and Remote Sensing 47 (2009) 722–738. [8] W.S. Lee, V. Alchanatis, C. Yang, M. Hirafuji, D. Moshou, C. Li, Sensing technologies for precision specialty crop production, Computers and Electronics in Agriculture 74 (2010) 2–33. [9] M. Montes-Hugo, S.C. Doney, H.W. Ducklow, W. Fraser, D. Martinson, S.E. Stammerjohn, O. Schofield, Recent changes in phytoplankton communities associated with rapid regional climate change along the Western Antarctic Peninsula, Science 323 (2009) 1470–1473. [10] K.S. Baker, R.C. Smith, Bio-optical classification and model of natural waters. 2, Limnology and Oceanography 27 (1982) 500–509. [11] M. Defoin-Platel, M. Chami, How ambiguous is the inverse problem of ocean color in coastal waters? Journal of Geophysical Research 112 (2007) C03004. [12] Z. Lee, R. Arnone, C. Hu, P.J. Werdell, B. Lubac, Uncertainties of optical parameters and their propagations in an analytical ocean color inversion algorithm, Applied Optics (2010) 369–381. [13] C.D. Mobley, L.K. Sundman, C.O. Davis, J.H. Bowles, T.V. Downes, R.A. Leathers, M.J. Montes, W.P. Bissett, D.D.R. Kohler, R.P. Reid, E.M. Louchard, A. Gleason, Interpretation of hyperspectral remote-sensing imagery by spectrum matching and lookup tables, Applied Optics 44 (2005) 3576–3592. [14] T.M. Mata, A.A. Martins, N.S. Caetano, Microalgae for biodiesel production and other applications: A review, Renewable and Sustainable Energy Reviews 14 (2010) 217–232. [15] K.J. Flynn, K. Davidson, A. Cunningham, Relations between carbon and nitrogen during growth of Nannochloropsis oculata (Droop) Hibberd under continuous illumination, New Phytologist 125 (1993) 717–722. [16] L.M. Lubián, O. Montero, I. Moreno-Garrido, I.E. Huertas, C. Sobrino, M. González-del Valle, G. Parés, Nannochloropsis (Eustigmatophyceae) as souce of commercially valuable pigments, Journal of Applied Phycology 12 (2000) 249–255. [17] E. Forján, I. Garbayo, C. Casal, C. Vílchez, Enhancement of carotenoid production in Nannochloropsis by phosphate and sulphur limitation, in: A. Méndez-Vilas (Ed.), Communicating Current Research and Educational Topics and Trends in Applied Microbiology, FORMATEX, Badajoz, Spain, 2007, pp. 356–364. [18] A. Solovchenko, I. Khozin-Goldberg, L. Recht, S. Boussiba, Stress-induced changes in optical properties, pigment and fatty acid content of Nannochloropsis sp.: Implications for non-destructive assay of total fatty acids, Marine Biotechnology 13 (2011) 527–535. [19] D. Pal, I. Khozin-Goldberg, Z. Cohen, S. Boussiba, The effect of light, salinity, and nitrogen availability on liped production by Nannochloropsis sp. Applied Microbiology and Biotechnology 90 (2011) 1429–1441.

[20] H.W. Paerl, J. Tucker, P.T. Bland, Carotenoid enhancement and its role in maintaining blue-green algal (Microcystis aeruginosa) surface blooms, Limnology and Oceanography 28 (1983) 847–857. [21] T. Berner, Z. Dubinsky, K. Wyman, P.G. Falkowski, Photoadaptation and the “package” effect in Dunaliella tertiolecta (Chlorophyceae), Journal of Phycology 25 (1989) 70–78. [22] R.A. Reynolds, D. Stramski, D.A. Kiefer, The effect of nitrogen limitation on the absorption and scattering properties of the marine diatom Thalassiosira pseudonana, Limnology and Oceanography 42 (1997) 881–892. [23] M.N. Merzlyak, O.B. Chivkunova, O.A. Gorelova, I.V. Reshetnikova, A.E. Solovchenko, I. Khozin-Goldberg, Z. Cohen, Effect of nitrogen starvation on optical properties, pigments, and arachidonic acid content of the unicellular green alga Parietochloris incisa (Trebouxiophyceae, Chlorophyta), Journal of Phycology 43 (2007) 833–843. [24] A.E. Solovchenko, I. Khozin-Goldberg, Z. Cohen, M.N. Merzlyak, Carotenoid-tochlorophyll ratio as a proxy for assay of total fatty acids and arachidonic acid content in the green microalgae Parietochloris incisa, Journal of Applied Phycology 21 (2009) 361–366. [25] Z. Lee, K.L. Carder, C.D. Mobley, R.G. Steward, J.S. Patch, Hyperspectral remote sensing for shallow waters I. A semianalytical model, Applied Optics 37 (1998) 6329–6338. [26] H.R. Gordon, O.B. Brown, M.M. Jacobs, Computed relationships between the inherent and apparent optical properties of a flat homogeneous ocean, Applied Optics 14 (1975) 417–427. [27] H.R. Gordon, O.B. Brown, R.H. Evans, J.W. Brown, R.C. Smith, K.S. Baker, D.K. Clark, A semianalytic radiance model of ocean color, Journal of Geophysical Research 93 (1988) 10,909–10,924. [28] Z. Lee, K.L. Carder, S.K. Hawes, R.G. Steward, T.G. Peacock, C.O. Davis, Model for the interpretation of hyperspectral remote-sensing reflectance, Applied Optics 33 (1994) 5721–5732. [29] R.W. Gould Jr., R.A. Arnone, Remote sensing estimates of inherent optical properties in a coastal environment, Remote Sensing of Environment 61 (1997) 290–301. [30] Z. Lee, K.L. Carder, C.D. Mobley, R.G. Steward, J.S. Patch, Hyperspectral remote sensing for shallow waters: 2. Deriving bottom depths and water properties by optimization, Applied Optics 38 (1999) 3831–3843. [31] K.L. Carder, F.R. Chen, Z.P. Lee, S.K. Hawes, D. Kamykowski, Semianalytic Moderate-Resolution Imaging Spectrometer algorithms for chlorophyll a and absorption with bio-optical domains based on nitrate-depletion temperatures, Journal of Geophysical Research 104 (1999) 5403–5421. [32] P. Forget, S. Ouillon, F. Lahet, P. Broche, Inversion of reflectance spectra of nonchlrophyllous turbid coastal waters, Remote Sensing of Environment 68 (1999) 264–272. [33] R.A. Reynolds, D. Stramski, B.G. Mitchell, A chlorophyll-dependent semianalytical reflectance model derived from field measurements of absorption and backscattering coefficients within the Southern Ocean, Journal of Geophysical Research 106 (2001) 7125–7138. [34] K.G. Ruddick, H.J. Gons, M. Rijkeboer, G. Tilstone, Optical remote sensing of chlorophyll a in case 2 waters by use of an adaptive two-band algorithm with optimal error properties, Applied Optics 40 (2001) 3575–3585. [35] T. Kutser, A. Herlevi, K. Kallio, H. Arst, A hyperspectral model for interpretation of passive optical remote sensing data from turbid lakes, Science of the Total Environment 268 (2001) 47–58. [36] Z. Lee, K.L. Carder, Effect of spectral band numbers on the retrieval of water column and bottom properties from ocean color data, Applied Optics 41 (2002) 2191–2201. [37] Z. Lee, K.L. Carder, R.A. Arnone, Deriving inherent optical properties from water color: a multiband quasi-analytical algorithm for optically deep waters, Applied Optics (2002) 5755–5772. [38] S. Maritorena, D.A. Siegel, A.R. Peterson, Optimization of a semianalytical ocean color model for global-scale applications, Applied Optics 41 (2002) 2705–2714. [39] C.S. Roesler, E. Boss, Spectral beam attenuation coefficient retrieved from ocean color inversion, Geophysical Research Letters 30 (2003) 21-1-4. [40] Z. Lee, K.L. Carder, Absorption spectrum of phytoplankton pigments derived from hyperspectral remote-sensing reflectance, Remote Sensing of Environment 89 (2004) 361–368. [41] K.L. Carder, F.R. Chen, J.P. Cannizzaro, J.W. Campbell, B.G. Mitchell, Performance of the MODIS semi-analytical ocean color algorithm for chlorophyll-a, Advances in Space Research 33 (2004) 1152–1159. [42] G. Dall'Olmo, A.A. Gitelson, Effect of bio-optical parameter variability on the remote estimation of chlorophyll-a concentration in turbid productive waters: experimental results, Applied Optics 44 (2005) 412–422 erratum 44 (2005) 3342. [43] D. Doxaran, N. Cherukuru, S.J. Lavender, Apparent and inherent optical properties of turbid estuarine waters: measurements, empirical quantification relationships, and modeling, Applied Optics 45 (2006) 2310–2324. [44] J.P. Cannizzaro, K.L. Carder, Estimating chlorophyll a concentrations from remotesensing reflectance in optically shallow waters, Remote Sensing of Environment 101 (2006) 13–24. [45] S.V. Salinas, C.W. Chang, S.C. Liew, Multiparameter retrieval of water optical properties from above-water remote-sensing reflectance using the simulated annealing algorithm, Applied Optics 46 (2007) 2727–2742. [46] C.-C. Liu, R.L. Miller, Spectrum matching method for estimating the chlorophyll-a concentration, CDOM ratio, and backscatter fraction from remote sensing of ocean color, Canadian Journal of Remote Sensing 34 (2008) 343–355. [47] Y. Oyama, B. Matsushita, T. Fukushima, K. Matsushige, A. Imai, Application of spectral decomposition algorithm for mapping water quality in a turbid lake (Lake Kasumigaura, Japan) from Landsat TM data, ISPRS Journal of Photogrammetry and Remote Sensing 64 (2009) 73–85.

T.A. Reichardt et al. / Algal Research 1 (2012) 22–31 [48] P. Shanmugam, B. Sundarabalan, Y.-H. Ahn, J.-H. Ryu, A new inversion model to retrieve the particulate backscattering in coastal/ocean waters, IEEE Transactions on Geoscience and Remote Sensing 49 (2011) 2463–2475. [49] A. Bricaud, A. Morel, Light attenuation and scattering by phytoplanktonic cells: a theoretical modeling, Applied Optics 25 (1986) 571–580. [50] D. Stramski, A. Morel, A. Bricaud, Modeling the light attenuation and scattering by spherical phytoplanktonic cells: a retrieval of the bulk refractive index, Applied Optics 27 (1988) 3954–3956. [51] S.E. Lohrenz, A.D. Weidemann, M. Tuel, Phytoplankton spectral absorption as influenced by community size structure and pigment composition, Journal of Plankton Research 25 (2003) 35–61. [52] K.R. Naqvi, T.H. Hassan, Y.A. Naqvi, Expeditious implementation of two new methods for analysing the pigment composition of photosynthetic specimens, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 60 (2004) 2783–2791. [53] T.G. Owens, J.C. Gallagher, R.S. Alberte, Photosynthetic light-harvesting function of violaxanthin in Nannochloropsis spp. (Eustigmatophyceae), Journal of Phycology 23 (1987) 79–85. [54] R.C. Smith, K.S. Baker, Optical properties of the clearest natural waters (200–800 nm), Applied Optics 20 (1981) 177–184. [55] F.E. Hoge, P.E. Lyon, Satellite retrieval of inherent optical properties by linear matrix inversion of oceanic radiance models: An analysis of model and radiance measurement errors, Journal of Geophysical Research 101 (1996) 16,631–16,648. [56] E.H.S. Van Duin, G. Blom, F.J. Los, R. Maffione, R. Zimmerman, C.F. Cerco, M. Dortch, E.P.H. Best, Modeling underwater light climate in relation to sedimentation, resuspension, water quality and autotrophic growth, Hydrobiologia 444 (2001) 25–42. [57] P. Latimer, E. Rabinowitch, Selective scattering of light by pigment-containing plant cells, Journal of Chemical Physics 24 (1956) 480. [58] A. Bricaud, A. Morel, L. Prieur, Optical efficiency factors of some phytoplankters, Limnology and Oceanography 28 (1983) 816–832. [59] R.J. Davies-Colley, R.D. Pridmore, J.E. Hewitt, Optical properties of some freshwater phytoplanktonic algae, Hydrobiologia 133 (1986) 165–178. [60] Y.-H. Ahn, A. Bricaud, A. Morel, Light backscattering efficiency and related properties of some phytoplankters, Deep Sea-Research 39 (11/12) (1992) 1835–1855. [61] F. Lahet, S. Ouillon, P. Forget, A three-component model of ocean color and its application in the Ebro River mouth area, Remote Sensing of Environment 72 (2000) 181–190. [62] M.N. Merzlyak, K.R. Naqvi, On recording the true absorption spectrum and scattering spectrum of a turbid sample: application to cell suspensions of the cyanobacterium Anabaena variabilis, Journal of Photochemistry and Photobiology. B 58 (2000) 123–129. [63] M.N. Merzlyak, O.B. Chivkunova, I.P. Maslova, K.R. Naqvi, A.E. Solovchenko, G.L. Klyachko-Gurvich, Light absorption and scattering by cell suspensions of some cyanobacteria and microalgae, Russian Journal of Plant Physiology 55 (2008) 420–425.

31

[64] C.D. Mobley, B. Gentili, H.R. Gordon, Z. Jin, G.W. Kattawar, A. Morel, P. Reinersman, K. Stamnes, R.H. Stavn, Comparison of numerical models for computing underwater light fields, Applied Optics 32 (1993) 7484–7504. [65] C.D. Mobley, Light and Water: Radiative Transfer in Natural Waters, Academic Press, New York, 1994. [66] S. Chandrasekhar, Radiative Transfer, Dover, New York, 1960. [67] J.T.O. Kirk, Volume scattering function, average cosines, and the underwater light field, Limnology and Oceanography 36 (1991) 455–467. [68] A. Morel, B. Gentili, Diffuse reflectance of oceanic waters: Bidirectional aspects, Applied Optics 32 (1993) 6864–6878. [69] N.K. Singh, S.G. Bajwa, I. Chaubey, Removal of surface reflection from above-water visible-near infrared spectroscopic measurements, Applied Spectroscopy 62 (2008) 1013–1021. [70] A. Gitelson, R. Stark, I. Dor, Quantitative near-surface remote sensing of wastewater quality in oxidation ponds and reservoirs: a case study of the Naan system, Water Environment Research 69 (1997) 1263–1271. [71] B. Fougnie, R. Frouin, P. Lecomte, P.-Y. Deschamps, Reduction of skylight reflection effects in the above-water measurement of diffuse marine reflectance, Applied Optics 38 (1999) 3844–3856. [72] A.A. Gitelson, J.F. Schalles, D.C. Rundquist, F.R. Schiebe, Y.Z. Yacobi, Comparative reflectance properties of algal cultures with manipulated densities, Journal of Applied Phycology 11 (1999) 345–354. [73] R.W. Gould Jr., R.A. Arnone, M. Sydor, Absorption, scattering, and remote-sensing reflectance relationships in coastal waters: Testing a new inversion algorithm, Journal of Coastal Research 17 (2001) 328–341. [74] H.R. Gordon, Diffuse reflectance of the ocean: the theory of its augmentation by chlorophyll a fluorescence at 685 nm, Applied Optics 18 (1979) 1161–1166. [75] S. Ahmed, A. Gilerson, J. Zhou, I. Ioannou, S. Hliang, B. Gross, F. Moshary, Impact of apparent fluorescence shift on retrieval algorithms for coastal waters, Proc. Ocean Optics XVIII, Montreal, Canada, 2006. [76] J.C. Lagarias, J.A. Reeds, M.H. Wright, P.E. Wright, Convergence properties of the Nelder-Mead simplex method in low dimensions, SIAM Journal on Optimization 9 (1998) 112–147. [77] A. Morel, In-water and remote measurements of ocean color, Boundary-Layer Meteorology 18 (1980) 177–201. [78] L. Prieur, S. Sathyendranath, An optical classification of coastal and oceanic waters based on the specific absorption curves of phytoplankton pigments, dissolved organic matter, and other particulate materials, Limnology and Oceanography 26 (1981) 671–689. [79] D. Haubrich, J. Musser, E.S. Fry, Instrumentation to measure the backscattering coefficient bb for arbitrary phase functions, Applied Optics 50 (2011) 4134–4147. [80] Z. Lee, Y.-H. Ahn, C. Mobley, R. Arnone, Removal of surface-reflected light for the measurement of remote-sensing reflectance from an above-surface platform, Optics Express 18 (2010) 26313–26324.