Open Access
Add to BibSonomy
Abstract
Quantum yield of fluorescence (ϕF) is key to interpret remote measurements of sun-induced fluorescence (SIF), and whether the SIF signal is governed by photochemical quenching (PQ) or non-photochemical quenching (NPQ). Disentangling PQ from NPQ allows using SIF estimates in various applications in aquatic optics. However, obtaining ϕF is challenging due to its high temporal and physiological variability, and the combined measurements needed to enclose all relevant optical paths. In inland waters, this type of data is scarce and information on diurnal and seasonal ϕF dynamics are almost unknown. Using an autonomous hyperspectral Thetis profiler in Lake Geneva, we demonstrate how to estimate ϕF using an ensemble of in-situ measurements acquired between 2018 to 2021. We use vertical and temporal changes in retrieved ϕF to determine NPQ and PQ conditions. We observed NPQ in 36% of the total daytime profiles used in the ϕF analysis. While downwelling irradiance is a significant contributor to ϕF, its role cannot be easily interpreted. Other factors such as phytoplankton photoregulation and assemblages also likely play significant roles in quenching mechanisms. We conclude that an adapted approach exploiting in-situ data is suitable to determine diurnal and seasonal NPQ occurrence, and helps develop future remote sensing algorithms.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Sun-induced chlorophyll fluorescence (SIF) emitted by chlorophyll-a (chl a) pigments in phytoplankton is widely used as a remote sensing (RS) proxy for chl a concentrations and related phytoplankton biomass [1–3] and phytoplankton physiology [4,5]. The SIF signal can be detected in the red to near-infrared (NIR) wavelength region of the electromagnetic spectrum with a peak at around 685 nm [6,7]. These wavelengths are much less affected by colored dissolved organic matter (CDOM) and non-algal particles (NAP) than shorter wavelengths [8], such as those used by blue-green algorithms. This puts SIF emission retrievals at an advantage for determining chl a concentration and ultimately, estimating phytoplankton biomass in optically complex waters. Further, SIF estimates are theoretically linked to the photosynthetic activity of phytoplankton communities in water bodies if the quenching mechanisms leading to SIF emission are known [4]. The complex interplay of processes determining SIF emissions, however, requires attention to accurately interpret and use SIF for phytoplankton research applications.
The additive contribution of SIF to the upwelling radiance (Lu) measured by a spectrometer can be described as [9]:
Photons absorbed in phytoplankton have three different fates: 1) charge separation in the reaction centers (RC) of Photosystem II (PS II) and Photosystem I (PS I) due to electron transfer, 2) SIF emission and 3) heat dissipation [10]. PS I and PS II are protein complexes where chl a pigments are bound to and absorb light with peaks roughly at 700 nm and 680 nm, respectively. In natural conditions, almost all phytoplankton SIF comes from PS II, making the influence of PS I negligible [10,11]. Furthermore, PS II SIF is the contribution observable at 685 nm [10]. PS I also contributes more to SIF emission at a longer wavelength region (∼740 nm for plants), where strong water absorption dominates light propagation in aquatic ecosystems [11].
The likelihood of the three different fates occurring depends on the combination of three factors; (1) the state of phytoplankton being dark- or light-regulated, (2) the closure or opening of RCs of PS II, and (3) the rate constants (k) of chl a de-excitation processes [9]. Fate 1 occurs in either dark- or light-regulated state when the RCs of PS II are open and undamaged. Most of the absorbed energy is used for charge separation in dark-regulated states, while the percentage varies in a light-regulated state, primarily due to ambient light. Fate 2 and 3 can occur in a dark- or light-regulated state, independent if RCs are open or closed. The main difference is that heat dissipation (Fate 3) is dominant during high incident light conditions [9].
SIF relates not only to chl a and phytoplankton biomass but also to photosynthetic activity of the phytoplankton community in the water body. Concerning the physiological component of SIF dynamics, a maximum fluorescence flux and subsequently maximum ${\phi _F}$ occurs under conditions when all RCs are closed. A reduction of ${\phi _F}$ due to increased photochemistry (i.e., Fate 1) leads to photochemical quenching (PQ). Another type of ${\phi _F}$ reduction not attributed to photochemistry (i.e., Fate 3) is called non-photochemical quenching (NPQ). Distinguishing the PQ and NPQ occurrences enables us to gain insights into the physiological underpinnings governing SIF dynamics detected via RS. However, identifying separately PQ and NPQ is difficult due to the complex biophysical mechanisms involved. For active fluorescence measurements, pulse amplitude modulation (PAM) or fast repetition rate (FRR) fluorometry have been used in many studies to understand fluorescence flux dynamics (e.g., [9,12,13]). In contrast, for passive fluorescence (i.e., SIF), where illumination conditions are uncontrolled, disentangling PQ from NPQ becomes even more challenging.
The complex variability of SIF due to PQ and NPQ interferences complicates the use of SIF for chl a concentration and phytoplankton biomass estimates. SIF-derived chl a concentration (assuming a linear relation between SIF and chl a) could, for example, show unexpected variability with lower chl a around solar noon and higher chl a at low light, while concentrations are not expected to change at comparable rates throughout the day [14]. A similar problem occurs when active fluorometers are used to derive chl a concentrations. Such sensors are therefore also susceptible to underestimation of chl a [14]. As outlined above, varying ${\phi _F}$ values provide information about PQ and NPQ dominant settings, and therefore phytoplankton physiology, and consequently, photosynthetic activity.
Several studies such as Maritorena et al. [14] and Morrison [11] for oceanic and Zhou et al. [15] for coastal waters have demonstrated how ${\phi _F}$ can be estimated and used to infer NPQ dominance. However, ${\phi _F}$ estimation is challenging especially in optically complex waters due to the input data necessary to calculate ${\phi _F}$ (i.e., SIF, ${E_O}({PAR} ),\; \; chl\; a,\; a_{phy.}^\ast ,\; {K_O}({PAR} ),\; \; {K_{Lu}}({\lambda _{em}}))$.
We argue that synchronous SIF and active fluorescence measurements at varying ${E_O}({PAR} )$ conditions provide insight into PQ and NPQ occurrences. We consequently developed a new method to determine NPQ using high-frequency measurements of apparent optical properties (AOP) and inherent optical properties (IOP) acquired by an automated profiler deployed in Lake Geneva since 2018 [15]. We calculated ${\phi _F}$ using automated in-situ data. We then distinguish NPQ and PQ occurrences from our ${\phi _F}$ results and evaluate the impact of NPQ and PQ dynamics on RS based SIF estimates.
2. Study site and data
2.1 LéXPLORE platform in Lake Geneva
Lake Geneva, the largest lake in Western Europe, is located on the border of Switzerland and France. This peri-alpine lake has a surface area of 580 km2 and a maximum depth of 310 m [16,17]. Several tributaries contribute to the lake’s water inflow with the Rhône River contributing 70%-75% [18]. Nutrient loadings, particularly of phosphorus, continuously increased from pre-industrial levels and led to cultural eutrophication of Lake Geneva in the 1970s [19]. A water management plan was successfully implemented to reduce phosphorus inputs and concentration [20]. However, phytoplankton biomass seemed to unexpectedly remain stable or slightly increase in the past two decades [21]. In 2018, the research platform LéXPLORE was installed in Lake Geneva to better understand biological, chemical and physical processes and how these are interrelated [22]. The platform is located 570 meters off the northern shore and towards the south-east of Lausanne, moored at 110 m depth (Fig. 1). The platform houses state-of-the-art instruments to continuously and simultaneously measure various physical and biological water and environmental parameters [22].
Fig. 1. (A) Lake Geneva showing the location of LéXPLORE (vector file from iStock.com/rbiedermann). (B) Thetis profiler enclosed in the grey box with LéXPLORE at the background. (C) Schematic of the profiler.
2.2 In-situ data
The Thetis autonomous profiler takes measurements in the top 50 m of the water column at an interval of 3 hours with a vertical resolution of 1 to 10 cm depending on the sensors’ sampling frequency and the profiling speed. We have obtained 1098 profiles between October 2018 and August 2021, half of which were measured during daytime. The Thetis profiler is moored at the lake bottom (Fig. 1(C)). An onboard electrical winch mounted on a positively buoyant frame allows for vertical profiling at desired time and speed. Thetis is equipped with probes and sensors which measure fluorescence, backscattering, absorption, attenuation, photosynthetically active radiation (PAR), upwelling radiance (Lu) and downwelling irradiance (Ed). All temperature-corrected in-situ measurements used in this study can be accessed in the Datalakes web portal (https://www.datalakes-eawag.ch/). A summary of the sensor and data information used herein is provided in Table 1. The phytoplankton absorption coefficient was obtained using the quantitative filter technique [24] in water samples taken between May and June 2021 since no water samples for $a_{phy}^\ast $ laboratory analysis were taken between 2018-2020.
Table 1. Measurements obtained from the Thetis profiler (specifications from https://www.seabird.com)
A correction and quality control procedure was undertaken for the data measured by the Spectral Absorption and Attenuation Meter (WetLabs AC-S). We applied a temperature correction, similar to Slade et al. (2010) [23], using the coefficients provided by the manufacturer and the water temperature from the CTD onboard the Thetis profiler. Furthermore, we used the M2 model from Pitarch et al. (2016) [24] to correct for the scattering error. M2 model removes the absorption residual at NIR wavelength (λ_NIR), but scales the error spectrally by the ratio of total scattering to the scattering at λ_NIR. λ _NIR was defined as the wavelength in the 690-715 [nm] range where measured absorption was the lowest [24]. A detailed description of the corrections applied are given in a supplement file in Minaudo et al. (2021) [15].
Shading corrections were considered negligible for this study. The Thetis profiler is anchored 30 m away from the LéXPLORE platform and calculations accounting for sun zenith angle showed that this would have no shading issues in our measurements. Furthermore, water visibility is not high.
3. Methods
3.1 SIF retrieval
We calculated SIF emission by using Thetis radiometer measurements and Hydrolight simulations [28] parameterized with Thetis IOP measurements. The measurements represent subsurface reflectance, including the additive SIF signal, while the model simulations were run with a ${\phi _F}$ of zero, and hence represent approximate subsurface reflectance excluding SIF. SIF was obtained as the difference between measured remote sensing reflectance ${R_m}$ and simulated remote sensing reflectance ${R_s}\; $ in the spectral interval between 660–700 nm (${\lambda _{em}}$) at depths of 0–30 m $(z )$ using the following equations,
All ${R_s}$ consider elastic scattering and Raman scattering, exclude SIF and were simulated with the Hydrolight radiative transfer model. For the ${R_s}$ simulations, Hydrolight was parameterized with simultaneously measured bulk absorption and scattering from the Thetis profiler. The Fournier-Forand phase function [25] together with the measured backscattering (bb) at four wavelengths (i.e., 440, 532, 630, and 700 [nm]) was used for each profile. The “semi-analytic” sky model was used wherein location and time were changed for each sampling period. Sun zenith angle changed for each simulation according to time. The sensitivity analysis showed a negligible effect of wind speed and cloud cover on the Rs spectrum for some randomly selected profiles. Therefore, a constant wind speed of 2 m/s (i.e., the annual wind speed locally observed) and clear sky conditions were assumed for those parameters.
In order to identify underestimations of CHLF due to NPQ, we calculated an absorption-based Chlorophyll concentration (CHLA) based on a technique used by Roesler and Barnard [14]. This procedure entails finding the linear relationship between night-time (low light) absorption line height (aLH) [26,27] and CHLF. This relationship is subsequently applied to daytime aLH to obtain CHLA. The aLH is calculated using Eqs. 4–5 [14].
For Eq. 2, a difference of maximally 10% between ${R_m}$ and ${R_s}\; $ outside the fluorescence emission region (i.e., 660 nm and 710 nm) was assigned as a threshold based on a previous study where optical closure in the red region resulted in a 10% average difference [28]. Such error metric was satisfied in only 17 profiles considering depths up to 15 m. We therefore used a bias correction approach [29], where a multiplicative coefficient is applied to ${R_s}\; $ such that the difference at the endpoints of the fluorescence emission region is minimized (see Supplement 1).
3.2 ${\phi _F}$ estimation
We calculate ${\phi _F}$ by adopting Eq. 6 as follows:
$SIF$ was obtained from calculations using Eq. 2; ${K_{Lu}}$ and ${K_O}$ from ${L_u}$ measurements and ${E_O}$ estimates (see Eqs. 8–(10)), respectively; ${\bar{a}_{phy}}$ (spectrally weighted ${a_{phy}})$ from the product of [chl a] and specific absorption coefficient of phytoplankton ($a_{phy}^\ast )$ and ${E_O}$, and $Q_a^\ast $ is estimated to be 0.5 based on literature [10]. ${K_{Lu}}$ is calculated by using two ${L_u}$ measurements at consecutive depths (${z_1}$ and ${z_2})$ from 0 to 40 m at 0.5 m interval according to [30,31] as:
We estimated ${E_O}$ through a three-step approach according to [29–31], the basis of which is Gershun’s equation, as:
Similar to ${K_{Lu}}$, we calculate ${K_d}$ by using two ${E_d}$ measurements at consecutive depths. ${K_d}$ was obtained using spectral ${E_d}$ data considering the PAR wavelength region. ${E_O}$ and K coefficients were calculated for each wavelength before they were summed for use in Eq. 6. It should be noted that the combination of Eqs. 8-10 to obtain ${E_O}$ is a simplified approach previously used in oceanic waters. Since we applied this calculation scheme for our oligo-mesotrophic study area, we verified that total spectral ${E_o}({PAR} )$ was comparable with the PAR sensor values (see Supplement 1).
To obtain ${\bar{a}_{phy}}$, we followed an approach presented in [11]:
The main sources of uncertainty in our dataset include low signal-to-noise ratio (SNR) of instruments at low light conditions and wave-focusing effects on near-surface Ed, both of which propagate through pre-processing steps for ${\phi _F}$ calculation. We set a limit to SNR based on sensor sensitivity values provided by the instrument manufacturer, which we used in uncertainty propagation calculations (see Supplement 1). Noisy and erratic input data near the water surface due to the influence of waves and strong winds were filtered and removed. The pre-processing resulted in 172 measured profiles that we used in the subsequent steps. The ${E_O}({PAR} )$, SIF and K coefficients were smoothened using Savitzky-Golay filter [32] along the depth with a window length of 5 data points at 0.5 m interval and polynomial order 2. From the uncertainty results, we set a threshold of depths up to 15 m for use in our analysis (see Supplement 1).
3.3 NPQ evaluation
We assess and quantify NPQ occurrence and dynamics using the sub-surface vertical dynamic of ${\phi _F}$. Assuming that SIF dynamics should follow the one of PAR under non-NPQ conditions, a reduction of ${\phi _F}$ (simply speaking the ratio between SIF and PAR) towards the surface would indicate the occurrence of NPQ. Despite the vertical heterogeneity in the water column, the dynamic [chl a] and attenuation are included in the estimates in ${\phi _F}$ which makes the aforementioned indication of NPQ occurrence valid. However, vertical non-uniformity due to varying phytoplankton species at different depths could potentially contribute to changes in ${\phi _F}$. The latter is not included in the study due to data unavailability
In particular, PQ conditions are apparent when ${\phi _F}$ increases with decreasing depth (i.e., increasing E(PAR)). This occurs since RCs tend to close when there is higher incident irradiance available. In contrast, NPQ becomes more dominant in cases where ${\phi _F}$ decreases with decreasing depth, typically approaching the shallowest part of the water column. This NPQ type can be linked to energy-dependent or photo-inhibition related quenching given the time scale of the interval between profiles [9]. This simple approach can be validated by analyzing the NPQ-corrected and uncorrected chl a profiles, CHLA and CHLF, respectively, as explained in Section 3.1. NPQ produces underestimation of fluorescence-based chl a measurements. Therefore, one can presume NPQ occurs in the region where uncorrected chl a is significantly lower than corrected chl a until the two corrected and uncorrected profiles intersect [14].
While Raman scattering is generally considered as contributor to radiances in oceanic waters especially at depth, this inelastic scattering is assumed negligible in our study: first, our analyses were limited to measurements at shallow depths [9,33]; second, Raman scattering contribution decreases significantly in the red wavelength region [34] where SIF is emitted; third, past studies have shown that their contribution is generally only significant for chl a < 1 mg m-3 [35], which only occurs in 2% of the dataset used in our study.
4. Results
4.1 SIF emission spectra
For the assessment of SIF spectra, we successfully processed 230 AOP/IOP profiles, roughly 20% of the original dataset. 395 suitable daytime profiles were initially used but 165 of the profiles were excluded due to failed optical closure: measured and simulated reflectances outside the fluorescence emission region indicated significant errors and biases. Despite applying the bias correction, the 10% threshold for the difference between ${R_m}$ and ${R_s}\; $ outside the fluorescence emission region (i.e., 660 nm and 710 nm) were not achieved by these 165 profiles. Additional plots showing examples of measured and simulated reflectances, and SIF spectra before and after applying a bias correction are shown in Supplement 1.
Overall, we observe a decreasing SIF emission with decreasing ${E_O}({PAR} )$ and increasing depth (Fig. 2(A)). This coincides with the proportionality between light intensity and SIF emission as shown in Eq. 1. Diurnal dynamics of SIF retrieved near the water surface are shown in Fig. 2(B).
Fig. 2. (A) A sample of sun-induced fluorescence (SIF) emission spectra retrieved on 11 April 2020 at 12:00 and (B) the range of SIF emission spectra retrieved closest to the water surface. The solid line and shaded region represent the mean and 95% confidence interval, respectively.
4.2 Vertical gradients of ${\phi _F}$ input profiles
A total of 172 input parameter profiles and resulting ${\phi _F}$ were analyzed, 58 less than the SIF spectra results. The relatively low number of used profiles is due to noisy input data originating from ${E_O}({PAR} ),\; \; {K_O}({PAR} )$ and ${K_{Lu}}({{\lambda_{em}}} )$ leading to high coefficient of variation in ${\phi _F}$ (see Supplement 1). In general, ${E_O}({PAR} )$ and SIF decrease with depth while ${\bar{a}_{phy}},$ ${K_O}({PAR} )$ and ${K_{Lu}}({{\lambda_{em}}} )$ profiles may oscillate due to varying concentrations of phytoplankton, suspended particulate and dissolved matter. Changes in input parameters and ${\phi _F}$ can be observed throughout the day (Fig. 3) and across seasons (Fig. 4). Typical vertical gradients are less pronounced for ${K_O}({PAR} )$ and ${K_{Lu}}({{\lambda_{em}}} )$, while chl a profiles either have a near- or subsurface chl a maximum [36]. Vertical shapes are diverse, either close to an exponential decay or a sigmoidal function with local maxima or minima deeper than 5 m, evidencing the complexity of vertical water IOP and constituent distributions.
Fig. 3. Profiles of photosynthetically active radiation (PAR), irradiance-weighted absorption coefficient of phytoplankton (${\bar{a}_{phy}}$), chlorophyll concentration (CHLA), sun-induced chlorophyll fluorescence (SIF), attenuation coefficients (Ko(PAR)+K(Lu)) and quantum yield (${\phi _F}$) obtained on 11 April 2020.
Fig. 4. Average seasonal profiles of photosynthetically active radiation (PAR), irradiance-weighted absorption coefficient of phytoplankton (${\bar{a}_{phy}}$), chlorophyll-a concentration (CHLA), sun-induced chlorophyll fluorescence (SIF), attenuation coefficients (K_o + K) and quantum yield (${\phi _F}$). Profiles were obtained near solar noon for (A) winter and spring and (B) summer and fall. Solid lines show mean values and the shaded region represents the standard deviation.
An example of a daily cycle of profiles from 11 April 2020 is shown in Fig. 3. The chl a profiles show changing phytoplankton dynamics throughout the day with sub-surface maximum depth increasing with increasing PAR. ${\bar{a}_{phy}}$ profiles show a similar sigmoidal shape as their chl a counterparts but with a generally decreasing trend related to the vertical decrease in PAR. The attenuation coefficient profiles vary significantly within the day, and show the influence of total absorption and scattering on PAR. Finally, we observe changing depths of maximum ${\phi _F}$. Profiles from 09h to 15h exhibit a sub-surface maximum while the maximum at 18h is at the shallowest point.
The seasonal cycles are grouped as follows: December-February, March-May, June-August, and September-November, corresponding to winter, spring, summer, and fall clusters, respectively. If profiles for specific seasons are available from different years, these were clustered still in one group due to little data availability for each season per year. The profiles shown in Fig. 4 represent the mean and distribution of profiles from 10h to 14h. The highest and lowest PAR values are observed in summer and winter, respectively. SIF is highest in summer and lowest in winter. Attenuation coefficients and ${\phi _F}$ generally decrease with depth for all seasons except summer.
4.3 Detection of NPQ occurrence
We exploited the depth at which maximum ${\phi _F}$ occurs in all profiles as a proxy for NPQ occurrence. In 62 measurements, the maximum in ${\phi _F}$ was not close to the surface but at deeper depths (as illustrated in Fig. 5), indicating an additional quenching of light energy, thus, the presence of NPQ. In other cases, the ${\phi _F}$ maximum was close to the surface, indicating NPQ was weak or absent (e.g., Fig. 6). We highlight the diurnal changes in ${\phi _F}$ maxima especially in Fig. 5, where the depth of maximum ${\phi _F}$ varies across the day and indicates increasing NPQ rates during high light conditions around noon and early afternoon in mid-April. Such changes are not visible in measurements from late October with lower irradiance (Fig. 6). The dynamic changes in maximum ${\phi _F}$ across the time period considered in our study are also plotted (see Supplement 1). ${\phi _F}$ values range from 0.001 to 0.02 with most data from 2020 and least in 2019.
Fig. 5. Diurnal changes in sun-induced chlorophyll fluorescence (SIF), downwelling scalar irradiance $({E_O}({PAR} )$ and quantum yield of fluorescence (${\phi _F})$ (top panel). The bottom panel show profiles of absorption-based (CHLA) and fluorometric chl a profiles (CHLF) with non-photochemical quenching (NPQ) occuring at 9 h, 12 h and 15 h. Panels from left to right represent increasing time of day on 11 April 2020.
Fig. 6. Diurnal changes in sun-induced chlorophyll fluorescence (SIF), downwelling scalar irradiance $({E_O}({PAR} )$ and quantum yield of fluorescence (${\phi _F})$ (top panel). The bottom panel shows absorption-based (CHLA) and fluorometric chl a profiles (CHLF) when non-photochemical quenching (NPQ) does not occur. Panels from left to right represent increasing time of day on 28 October 2018.
Comparing absorption-based CHLA and fluorometric CHLF profiles additionally supports the aforementioned findings (i.e., green bold and dashed lines in Fig. 5 and Fig. 6). We particularly noted cases where CHLF values are significantly lower (> 20% difference) than CHLA near the water surface. The difference between CHLA and CHLF diminishes with depth and at a certain depth, both variables are hardly distinguishable. Such scenarios indicate cases of large underestimations in CHLF, and an evidence of the NPQ occurrence. These observations align with the previously presented ${\phi _F}$ maxima analysis. In Fig. 5, for example, large differences in CHLA and CHLF are found during the same hours as when ${\phi _F}$ reaches a deep maximum. Although the depths showing extent of NPQ from the ${\phi _F}$ and CHLF-CHLA do not exactly match, we can use both approaches to approximate the depths affected by NPQ. In Fig. 6, we do not see differences between CHLA and CHLF nor a sub-surface ${\phi _F}$ peak, indicating no significant NPQ occurrence.
The same two modalities found for changes in ${\phi _F}$ with depth are observed for the dependency of ${\phi _F}$ on ${E_O}({PAR} )$: ${\phi _F}$ was either continuously increasing with ${E_O}({PAR} )$, or peaked and declined under higher levels of irradiance. The first modality is the result of a decreasing photosynthetic efficiency with increasing light availability, which causes an increase in SIF. The second modality indicates NPQ occurrence at the peaking position as a result of the phytoplankton community reaching its ${E_O}({PAR} )$ saturation threshold (Fig. 7).
Fig. 7. Quantum yield of fluorescence (${\phi _F})$ against downwelling irradiance (${E_O}({PAR} )$). (A) Values from 11 April 2020 showing non-photochemical quenching (NPQ) occurrence between 9h-15 h and (B) values from 28 October 2019 showing no NPQ throughout the day.
Aside from the diurnal variability, we investigated potential trends in ${\phi _F}$ by clustering daytime dataset between seasons. Related seasonal variability between ${\phi _F}$ and ${E_O}({PAR} )$ is shown in Fig. 8. A monotonous increase of ${\phi _F}$ with increasing ${E_O}({PAR} )$ for 110 instances across all seasons denoting PQ, is shown in Fig. 8. The NPQ case shows a peak at 100 umol m-2 s-1 for spring and summer for 62 instances but was not observed in data from winter and fall.
Fig. 8. Quantum fluorescence (${\phi _F})$ against downwelling irradiance (${E_O}({PAR} )$ between different seasons. Solid lines and shaded regions indicate the mean and the 95% confidence interval, respectively. Photochemical quenching (PQ) cases show an increasing trend for all seasons, while non-photochemical quenching (NPQ) cases occurring in spring and summer with an inset of spring plot is shown to emphasize the inflection.
NPQ caused an underestimation of up to 75% in chl a (CHLF) in the top 2 m of the water column under the highest irradiance levels (Fig. 9). In NPQ cases, the relative difference in topmost and subsurface maximum of ${\phi _F}$ reaches about 80%. There is no clear trend of how chl a concentration affects the relative difference of CHLA and CHLF, but we observe an increasing trend for the difference between CHLA and CHLF estimates with PAR. In a quarter of the data, overestimated chl a values (between ∼1-20%) correspond to instances where we have subsurface ${\phi _F}$ maximum. This implies that in some cases, these two approaches don’t coincide in determining NPQ instances.
Fig. 9. Relative difference of absorption-based chl a (CHLA) and fluorometric chl a (CHLF) against irradiance in the upper 2 m of the water column for all non-photochemical quenching (NPQ) cases. Circle sizes depict the level of near-surface absorption-based CHLA (mg m-3).
5. Discussion
5.1 Suitability of Thetis profiler for determining fluorescence quenching
Our study is the first demonstration of using the sensor ensemble from a Thetis profiler to calculate SIF, estimate ${\phi _F}$ and assess NPQ occurrences in inland waters. While previous studies investigated ${\phi _F}$ and quenching in SIF emission [1,11,37], their approach relied on modelled SIF, and investigations were done in oceanic waters with phytoplankton and covarying material as the only assumed contributor to attenuation. Despite the availability of a relatively large volume of in-situ data, we encountered two main challenges resulting in a substantial reduction of safely usable measurements for our analysis. One critical aspect is that the required optical closure to retrieve SIF spectra relies heavily on the quality of IOP and AOP measurements (Sec. 5.2). Another difficulty is to obtain ${E_O}({PAR} )$ and ${K_O}({PAR} )$ from their planar counterparts due to the anisotropy of the underwater light field. Tackling both challenges requires careful data processing and filtering to avoid misinterpretation of subsequently derived results. Our approach shows that with the available sensors and applied data processing, it is possible to reliably retrieve SIF and ${\phi _F}$ occurrence in inland waters (Lake Geneva in our case) for 21% and 16%, respectively, of all automatically acquired profiles.
5.2 SIF retrieval considerations
We demonstrated a measurement-based approach complemented by radiative transfer modelling to retrieve SIF from spectral reflectance in inland waters. This retrieval method has advantages compared to the widely-used Fluorescence Line Height (FLH). FLH has many assumptions which fail in many water types, specifically in optically complex waters such as our study area [38]. Many studies also correlate FLH with chl a (e.g., [39,40]). In our study, we used an approach to estimate SIF independent of chl a and rather with an application of gaining more understanding of photosynthetic activity in mind. Several previous studies [1,11,37] opted to apply a more sophisticated approach by modelling a Gaussian shaped spectrum using the upwelling radiance at the SIF emission peak (i.e., 685 nm) to eventually retrieve SIF. While this might be a better representation of the SIF contribution to the upwelling radiance signal compared to FLH, this technique assumes a uniform Gaussian shape for all cases. Such an approach is standard in SIF remote sensing research, however, with increased availability of hyperspectral AOP and IOP data, we followed a measurement-based method that would characterize a realistic SIF spectrum with less initial assumptions. The main limitations of our approach are on one hand that it requires dual instrumental closure to obtain independent AOP and IOP measurements, and it is hence not transferable to above-surface sensors. On the other hand, high uncertainty in AOP and IOP measurements, in particular the backscattering coefficients, requires filtering of relatively many observations that are subject to poor optical closure [33]. Nevertheless, the bias correction implemented to address the optical closure issue resulted in 230 SIF spectra estimates. A comparison of Gaussian shape and measurement-based SIF spectra calculation is not included in this work but could be addressed in more detail in a future study.
5.3 ${\phi _F}$ calculation and vertical gradient
Temporal variability of ${\phi _F}$ and its driving parameters during the day and between seasons demonstrate the large dynamic nature of biophysical variables in the water column. Changes in ${\phi _F}$ in the water column during different measurement periods have been demonstrated in past studies in oceanic waters [11,37]. Such variabilities indicate the importance but also challenges of calculating ${\phi _F}$ for optically complex waters.
During the day, ${E_O}({PAR} )$ and SIF mainly determine the decreasing ${\phi _F}\; $ values with depth whereas K and ${\bar{a}_{phy}}$ (including chl a used as input for ${\bar{a}_{phy}}$) significantly impact ${\phi _F}$ peak values at depth. In general, we see a strong inverse relationship between ${\bar{a}_{phy}}$ and ${\phi _F}$ in cases where there is a sub-surface ${\phi _F}$ peak. This can already indicate which quenching mechanism (PQ vs NPQ) is dominant during the measurement period and is further discussed in Section 5.5. We acknowledge diel patterns in water optical properties and water constituents can be due to physical processes such as internal motion, mixing/restratification diel cycles, or lateral advection of different water masses [15,41]. It remains challenging to quantify the influence of these physical processes on the diel signals observed in the water optical properties. Therefore, their possible influence was not considered in our study.
The seasonal profiles provide information on intra-annual changes of ${\phi _F}$ and its driving variables. What is distinct for summer profiles is the sub-surface inflection in ${\phi _F}$, which contrasts with generally decreasing ${\phi _F}$ in all other seasons. While this provides a general impression of ${\phi _F}$ dynamics within each season, this observation does not imply that other ${\phi _F}$ vertical shapes cannot occur as demonstrated by a sub-surface ${\phi _F}$ peak in Fig. 5, which was observed during spring. These seasonal profiles merely provide the dominant profile features from our dataset. This may be an indication of a transition period either due to biotic or abiotic factors which may lead to different phytoplankton photo-adaptive measures.
The ${\phi _F}$ values calculated are generally lower than previous studies done in oceanic waters [11,37] but in a similar range with coastal waters [42]. The lower ${\phi _F}$ values compared to oceanic waters could be due to more absorbed light being used for photochemistry and therefore primary productivity [43]. The low ${\phi _F}$ values may also be associated with uncertainty in $Q_a^\ast $, which is currently assumed to be constant at 0.5 based on literature since there is currently no method which can measure this outside a laboratory setting [42].
5.4 NPQ evaluation and interpretation
In the profiles where ${\phi _F}$ increases with irradiance, PQ decreases due to a decrease in open RCs [44]. This light emission is a passive energy pathway allowing phytoplankton to dissipate the absorbed incident light energy when RCs are partly closed and too much energy is available [11,44]. Such cases are shown in Fig. 5 (18h, profile, and below the inflection point in 09h to 15h profiles) and in Fig. 6. In cases when ${\phi _F}$ decreases with irradiance, NPQ increases potentially due to energy-dependent quenching (qE) as shown in Fig. 5 (above the inflection point in 09h to 15h). Under high irradiance, a build-up of the electrochemical gradient in the thylakoid membrane occurs leading to qE. This type of quenching is an active, rapid and efficient way of managing excessive energy in RCs when irradiance is too high. NPQ may also increase due to decreasing PSII absorption cross-sections (qT) or damage in photosynthetic apparatus (qI) undergone by the phytoplankton [9,11]. At this stage, it is difficult to distinguish which type of NPQ mechanism is in place but some studies [9,11] indicate a time scale of seconds to a few minutes (less than 20 minutes), and minutes to hours for qE, qT, and qI, respectively. In the profiles where we observe NPQ, the difference in ${\phi _F}$ between the shallowest depth and sub-surface maximum can vary between 30-70%. This range coincides with previous studies wherein a decrease in ${\phi _F}$ of up to 90% and 40% were observed in qE and qI for marine phytoplankton [9,45].
The chl a profiles show concurrent quenching occurrences with the ones indicated by ${\phi _F}$ profiles. The depths at which CHLF and CHLA overlap, and the ones at which we observe a sub-surface maximum of ${\phi _F}$ coincide at ± 2 m. Differences between the two approaches in the identification of the depth where quenching shifts from NPQ to PQ can be attributed to inherent differences in SIF and active fluorometry processes. The uncertainties in ${\phi _F}$ calculations also play a role, given the uncertainties of the various inputs. Nevertheless, the general characteristics of PQ and NPQ occurrences match in 75% of these two profile types. We did not observe any correlation between the relative changes of CHLF and CHLA in near-surface chl a and the difference between near-surface and subsurface maximum ${\phi _F}$. This can be attributed to the CHLF and CHLA difference being at the same depth while the ${\phi _F}$ difference describes varying quantum efficiency at different depths. A better comparison would be having both near-surface ${\phi _F}$ estimates available when NPQ is activated and when it is minimal, but this is impossible to retrieve under natural illumination.
The increase in ${\phi _F}$ due to a decrease in PQ, and the decrease in ${\phi _F}$ due to an increase in NPQ can both be attributed to phytoplankton dynamics under changing and increasing light availability. The main difference between the two is the magnitude of light available and the light exposure history of phytoplankton. NPQ dominates when phytoplankton communities are subjected to higher irradiances and/or exposed to prolonged excessive light [14,46]. In Fig. 7(A), there is an increase in ${\phi _F}$ along with irradiance until ${E_O}({PAR} )$ value reaches roughly 102 umol m-2 s-1, which may indicate a saturation irradiance for photosynthesis [14]. This value coincides with the ${E_O}({PAR} )$ value at which inflection point is also observed in Morisson’s study [11]. However, looking at the seasonal plots of ${\phi _F}$ against ${E_O}({PAR} )$ (Fig. 8, PQ case), a general trend of increasing ${\phi _F}$ with irradiance is shown beyond ${E_O}({PAR} )$ of 103 umol m-2 s-1, whereas the NPQ case exhibits an inflection point at 102 umol m-2 s-1 for spring and summer. This shows that there is a critical point where saturation irradiance is reached at ∼102 umol m-2 s-1 when NPQ is activated in many yet not all instances. This difference can be attributed to light exposure history and phytoplankton assembly. Spring has a flatter slope for the NPQ trend which may be an indication of a seasonal or environmental transition point leading to more pronounced inflection point in summer. We do not know the phytoplankton communities represented in our data, which currently limits further investigation. The water column stratification can also play a role in driving vertical heterogeneity of species composition, and consequently vertical variation of ${\phi _F}$. In contrast, one can assume relatively uniform species composition under weak stratification (i.e., deep mixed-layer depth) conditions. However, investigating the effect of water column stratification is beyond the scope of this study.
5.5 Implications of ${\phi _F}$ and quenching mechanisms to understand SIF dynamics
SIF retrievals are typically used as proxy or input variable for estimates of chl a concentration or determining the photosynthetic activity of phytoplankton using Earth Observation approaches. However, dynamics in ${\phi _F}$ due to quenching mechanisms and, thus, resulting dynamics in SIF challenges the use of retrieved SIF to simultaneously estimate biomass and primary production. While biomass estimates rely on potential SIF with a maximum and constant ${\phi _F}$, the production estimates require actual SIF with varying ${\phi _F}$ as input [11]. Our results indicate the potential in generating a compensation scheme for underestimated SIF-based chl a measurements due to changing ${\phi _F}$ but such empirical schemes will likely be regional and time sensitive. Using ${\phi _F}$ in primary production models is more plausible since we can estimate carbon assimilation using a simple model requiring SIF, especially if we are able to obtain attenuation coefficients and quantum yield of carbon fixation as input [9].
Retrieving and interpreting SIF from remote sensing data is challenging especially from sun-synchronous orbiting satellite missions. Present knowledge on ${\phi _F}$ dynamics and subsequent quenching occurrences from in-situ data need to be scaled up to airborne and spaceborne data. For instance, knowing the PAR available at the time of a satellite overpass coupled with the understanding of phytoplankton SIF dynamics in the target water body are essential in applying SIF for both chl a estimates and primary productivity modelling. Our findings suggest a potential to assimilate such information in remote sensing algorithms. As an example, we can use the saturation irradiance obtained from our results as a threshold in determining whether SIF retrieved is emitted during an NPQ-dominated case. Almost half of our observations in spring and summer exhibit the ∼102 umol m-2 s-1 PAR threshold. This will enable us to discern whether the SIF and the corresponding ${\phi _F}$ obtained is primarily due to heat dissipation (NPQ case) or photosynthesis (PQ).
Several recent and upcoming satellite missions such as ESA’s Fluorescence Explorer (FLEX), ASI’s Hyperspectral PRecursor of the Application Mission (PRISMA), DLR’s Environmental Mapping and Analysis Program (EnMAP) and NASA’s Phytoplankton, Aerosol, Cloud, ocean Ecosystem (PACE) have a hyperspectral resolution which can enable a robust retrieval of SIF and potentially ${\phi _F}$ [38].The FLEX mission has the main objectives of monitoring chlorophyll fluorescence, photosynthetic activity and vegetation stress in terrestrial environment [47]. Our study contributes to demonstrating the potential applicability of FLEX to aquatic environments and SIF signal interpretation.
6. Conclusion
Our results provide insights into the relationship between SIF emission, ${\phi _F}$ and quenching mechanisms in inland waters. Particularly ${\phi _F}\; $ is a small quantity, challenging to calculate, and yet contains information on quenching mechanisms crucial for a meaningful interpretation of SIF. We conclude on large dynamic changes in ${\phi _F}$ as a function of varying environmental drivers and the composition of PQ and NPQ along the water column, throughout the day and between seasons. NPQ occurs in 36% of the total profiles analyzed, diminishing ${\phi _F}$ by as much as 30% with respect to the peak ${\phi _F}$ calculated within the same measurement period. In presence of NPQ, we found CHLF to underestimate CHLA by up to 75%. We conclude that the interpretation of SIF-based remote sensing approaches must consider contrasting quenching conditions. More in-depth investigations on environmental and ecological variables contributing to changes in SIF emission, ${\phi _F}$ and quenching is necessary to identify specific factors which significantly impact phytoplankton photoadaptation and regulation and eventually elaborate strategies to disentangling NPQ from PQ events and occurrences.
Our results suggest a saturation irradiance of photosynthesis at a diurnal and seasonal time scale for spring and summer of roughly 102 umol m-2 s-1, coinciding with saturation irradiance found in previous ${\phi _F}$ studies in marine environments [11,37]. We found that high incident irradiance and light exposure history are the main drivers of quenching mechanisms, but other variables such as phytoplankton communities may also significantly impact the occurrence of quenching processes. We suggest further analysis to accurately pinpoint the causes of this saturation irradiance since it does not occur in all cases investigated
We recommend regular monitoring of phytoplankton groups via an automated flow cytometry or an underwater imager and taking profiles at a higher temporal resolution to provide more insight into changing phytoplankton dynamic and its dependence on prolonged light exposure at varying magnitude. This will allow us to accurately record the onset of changing quenching mechanisms, especially NPQ, which have subtypes triggered at varying timescales due to different mechanisms within the phytoplankton cell. Furthermore, we recommend applying the same approach to other water bodies where in-situ measurements are available. Since ${\phi _F}$ and quenching properties in phytoplankton is an understudied field of research, we need to build knowledge in the inland water community to assess how regional or universal the findings are from Lake Geneva. Such knowledge building also makes SIF retrieval from satellite data more explainable.
Funding
Swiss Federal Office of the Environment (200021_179123).
Acknowledgments
The authors would like to thank Anita Schlatter, Pascal Rünzi, Charis Gretler, and Sébastien Lavanchy for assisting in the fieldwork and laboratory measurements. We thank the LéXPLORE platform team for their administrative and technical support and for the LéXPLORE core dataset. We also acknowledge the five LéXPLORE partner institutions: Eawag, EPFL, University of Geneva, University of Lausanne, and CARRTEL (INRAE-USMB). We also thank Emmanuel Boss and the three anonymous reviewers for their constructive feedback and comments.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are available from the Datalakes website [48].
Supplemental document
See Supplement 1 for supporting content.
References
1. A. Gilerson, J. Zhou, S. Hlaing, I. Ioannou, J. Schalles, B. Gross, F. Moshary, and S. Ahmed, “Fluorescence component in the reflectance spectra from coastal waters. Dependence on water composition,” Opt. Express 15(24), 15702–15721 (2007). [CrossRef]
2. H. J. Gons, M. T. Auer, and S. W. Effler, “MERIS satellite chlorophyll mapping of oligotrophic and eutrophic waters in the Laurentian Great Lakes,” Remote Sens. Environ. 112(11), 4098–4106 (2008). [CrossRef]
3. J. Gower and S. King, “Validation of chlorophyll fluorescence derived from MERIS on the west coast of Canada,” Int. J. Remote Sens. 28(3-4), 625–635 (2007). [CrossRef]
4. M. J. Behrenfeld, T. K. Westberry, E. S. Boss, R. T. O’Malley, D. A. Siegel, J. D. Wiggert, B. A. Franz, C. R. Mcclain, G. C. Feldman, S. C. Doney, J. K. Moore, G. Dall’olmo, A. J. Milligan, I. Lima, and N. Mahowald, “Satellite-detected fluorescence reveals global physiology of ocean phytoplankton,” Biogeosciences 6(5), 779–794 (2009). [CrossRef]
5. Y. Huot, B. A. Franz, and M. Fradette, “Estimating variability in the quantum yield of Sun-induced chlorophyll fluorescence: A global analysis of oceanic waters,” Remote Sens. Environ. 132, 238–253 (2013). [CrossRef]
6. R. A. Neville and J. F. R. Gower, “Passive remote sensing of phytoplankton via chlorophyll a fluorescence,” J. Geophys. Res. 82(24), 3487–3493 (1977). [CrossRef]
7. H. R. Gordon, “Diffuse reflectance of the ocean: the theory of its augmentation by chlorophyll a fluorescence at 685 nm,” Appl. Opt. 18(8), 1161–1166 (1979). [CrossRef]
8. A. A. Gilerson, A. A. Gitelson, J. Zhou, D. Gurlin, W. Moses, I. Ioannou, and S. A. Ahmed, “Algorithms for remote estimation of chlorophyll-a in coastal and inland waters using red and near infrared bands,” Opt. Express 18(23), 24109–24125 (2010). [CrossRef]
9. Y. Huot and M. Babin, “Overview of Fluorescence Protocols: Theory, Basic Concepts, and Practice,” in Chlorophyll a Fluorescence in Aquatic Sciences: Methods and Applications (Springer Netherlands, 2010), pp. 31–74.
10. P. Falkowski and D. A. Kiefer, “Chlorophyll a fluorescence in phytoplankton: Relationship to photosynthesis and biomass,” J. Plankton Res. 7(5), 715–731 (1985). [CrossRef]
11. 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(2), 618–631 (2003). [CrossRef]
12. N. Schuback, P. D. Tortell, I. Berman-Frank, et al., “Single-Turnover Variable Chlorophyll Fluorescence as a Tool for Assessing Phytoplankton Photosynthesis and Primary Productivity: Opportunities, Caveats and Recommendations,” Front. Mar. Sci. 8, 690607 (2021). [CrossRef]
13. M. A. Lucius, K. E. Johnston, L. W. Eichler, J. L. Farrell, V. W. Moriarty, and R. A. Relyea, “Using machine learning to correct for nonphotochemical quenching in high-frequency, in vivo fluorometer data,” Limnol. Oceanogr.: Methods 18(9), 477–494 (2020). [CrossRef]
14. C. S. Roesler and A. H. Barnard, “Optical proxy for phytoplankton biomass in the absence of photophysiology: Rethinking the absorption line height,” Methods in Oceanography 7, 79–94 (2013). [CrossRef]
15. C. Minaudo, D. Odermatt, D. Bouffard, A. I. Rahaghi, S. Lavanchy, and A. Wüest, “The Imprint of Primary Production on High-Frequency Profiles of Lake Optical Properties,” Environ. Sci. Technol. 55(20), 14234–14244 (2021). [CrossRef]
16. D. Bouffard and M. E. Perga, “Are flood-driven turbidity currents hot spots for priming effect in lakes?” Biogeosciences 13(12), 3573–3584 (2016). [CrossRef]
17. I. Kiefer, D. Odermatt, O. Anneville, A. Wüest, and D. Bouffard, “Application of remote sensing for the optimization of in-situ sampling for monitoring of phytoplankton abundance in a large lake,” Sci. Total Environ. 527-528(September 2015), 493–506 (2015). [CrossRef]
18. F. Rapin and D. Gerdeaux, “La protection du Léman priorité à la lutte contre l’eutrophisation,” Arch Sci 66(2), 103–116 (2013).
19. F. Rapin, P. Blanc, and C. Corvi, “Influence des apports sur le stock de phosphore dans le Lac Leman et sur son eutrophisation,” Revue des Sciences de l’Eau 2(4), 721–737 (2005). [CrossRef]
20. O. Anneville and J. P. Pelletier, “Recovery of Lake Geneva from eutrophication: Quantitative response of phytoplankton,” Arch Hydrobiol 148(4), 607–624 (2000). [CrossRef]
21. B. Müller, T. Steinsberger, A. Stöckli, and A. Wüest, “Increasing Carbon-to-Phosphorus Ratio (C:P) from Seston as a Prime Indicator for the Initiation of Lake Reoligotrophication,” Environ. Sci. Technol. 55(9), 6459–6466 (2021). [CrossRef]
22. A. Wüest, D. Bouffard, J. Guillard, B. W. Ibelings, S. Lavanchy, M. E. Perga, and N. Pasche, “LéXPLORE: A floating laboratory on Lake Geneva offering unique lake research opportunities,” Wiley Interdiscip. Rev.: Water 8(5), 1–15 (2021). [CrossRef]
23. W. H. Slade, E. Boss, G. Dall’olmo, M. R. Langner, J. Loftin, M. J. Behrenfeld, C. Roesler, and T. K. Westberry, “Underway and moored methods for improving accuracy in measurement of spectral particulate absorption and attenuation,” J Atmos Ocean Technol 27(10), 1733–1746 (2010). [CrossRef]
24. J. Pitarch, G. Volpe, S. Colella, R. Santoleri, and V. Brando, “Absorption correction and phase function shape effects on the closure of apparent optical properties,” Appl. Opt. 55(30), 8618–8636 (2016). [CrossRef]
25. C. D. Mobley, L. K. Sundman, and E. Boss, “Phase function effects on oceanic light fields,” Appl. Opt. 41(6), 1035–1050 (2002). [CrossRef]
26. E. S. Boss, R. Collier, G. Larson, K. Fennel, and W. S. Pegau, “Measurements of spectral optical properties and their relation to biogeochemical variables and processes in Crater Lake, Crater Lake National Park, OR,” Hydrobiologia 574(1), 149–159 (2007). [CrossRef]
27. R. F. Davis, C. C. Moore, J. R. V. Zaneveld, and J. M. Napp, “Reducing the effects of fouling on chlorophyll estimates derived from long-term deployments of optical instruments,” J. Geophys. Res. 102(C3), 5851–5855 (1997). [CrossRef]
28. D. M. O’Donnell, S. W. Effler, C. M. Strait, and G. A. Leshkevich, “Optical characterizations and pursuit of optical closure for the western basin of Lake Erie through in situ measurements,” J. Great Lakes Res. 36(4), 736–746 (2010). [CrossRef]
29. Y. Tian, G. J. Huffman, R. F. Adler, L. Tang, M. Sapiano, V. Maggioni, and H. Wu, “Modeling errors in daily precipitation measurements: Additive or multiplicative?” Geophys. Res. Lett. 40(10), 2060–2065 (2013). [CrossRef]
30. K. J. Voss, H. R. Gordon, S. Flora, B. Carol Johnson, M. Yarbrough, M. Feinholz, and T. Houlihan, “A Method to extrapolate the diffuse upwelling radiance attenuation coefficient to the surface as applied to the Marine Optical Buoy (MOBY),” J Atmos Ocean Technol 34(7), 1423–1432 (2017). [CrossRef]
31. H. Arst, S. Mäekivi, T. Lukk, and A. Herlevi, “Calculating irradiance penetration into water bodies from the measured beam attenuation coefficient,” Limnol. Oceanogr. 42(2), 379–385 (1997). [CrossRef]
32. A. Savitzky and M. J. E. Golay, “Smoothing and Differentiation of Data by Simplified Least Squares Procedures,” Anal. Chem. 36(8), 1627–1639 (1964). [CrossRef]
33. R. Frouin, P.-Y. Deschamps, and P. Dubuisson, “Remote sensing of solar-stimulated phytoplankton chlorophyll absorption in the oxygen B-band,” Proc. SPIE 7150(December 2008), 715005 (2008). [CrossRef]
34. J. von Bismarck and J. Fischer, “An examination of errors in computed water-leaving radiances due to a simplified treatment of water Raman scattering effects,” AIP Conf Proc 1531(May), 939–942 (2013). [CrossRef]
35. A. Morel, D. Antoine, and B. Gentili, “Bidirectional reflectance of oceanic waters: accounting for Raman emission and varying particle scattering phase function,” Appl. Opt. 41(30), 6289 (2002). [CrossRef]
36. V. Nouchi, D. Odermatt, A. Wüest, and D. Bouffard, “Effects of non-uniform vertical constituent profiles on remote sensing reflectance of oligo- to mesotrophic lakes,” Eur. J. Remote Sens. 51(1), 808–821 (2018). [CrossRef]
37. S. Maritorena, A. Morel, B. Gentili, S. Maritorena, A. Morel, and B. Gentili, “Determination of the fluorescence quantum yield by oceanic phytoplankton in their natural habitat,” Appl. Opt. 39(36), 6725–6737 (2000). [CrossRef]
38. R. S. Gupana, D. Odermatt, I. Cesana, C. Giardino, L. Nedbal, and A. Damm, “Remote Sensing of Environment Remote sensing of sun-induced chlorophyll- a fluorescence in inland and coastal waters : Current state and future prospects,” Remote Sens. Environ. 262(April), 112482 (2021). [CrossRef]
39. H. J. Gons, “A chlorophyll-retrieval algorithm for satellite imagery (Medium Resolution Imaging Spectrometer) of inland and coastal waters,” J. Plankton Res. 24(9), 947–951 (2002). [CrossRef]
40. J. Gower and S. King, “Use of satellite images of chlorophyll fluorescence to monitor the spring bloom in coastal waters,” Int. J. Remote Sens. 33(23), 7469–7481 (2012). [CrossRef]
41. B. Fernández Castro, H. E. Chmiel, C. Minaudo, S. Krishna, P. Perolo, S. Rasconi, and A. Wüest, “Primary and Net Ecosystem Production in a Large Lake Diagnosed From High-Resolution Oxygen Measurements,” Water Resour. Res. 57(5), e2020WR029283 (2021). [CrossRef]
42. J. Zhou, A. Gilerson, I. Ioannou, S. Hlaing, J. Schalles, B. Gross, F. Moshary, and S. Ahmed, “Retrieving quantum yield of sun-induced chlorophyll fluorescence near surface from hyperspectral in-situ measurement in productive water,” Opt. Express 16(22), 17468 (2008). [CrossRef]
43. D. C. Pierson, “Light and Primary Production in Lakes,” in Encyclopedia of Lakes and Reservoirs, R. W. and F. W R. Bengtsson Lars and Herschy, ed. (Springer Netherlands, 2012), pp. 485–492.
44. D. A. Kiefer and R. A. Reynolds, “Advances in Understanding Phytoplankton Fluorescence and Photosynthesis,” Primary Productivity and Biogeochemical Cycles in the Sea 3, 155–174 (1992). [CrossRef]
45. A. v Ruban, J. Lavaud, B. Rousseau, G. Guglielmi, P. Horton, and A.-L. Etienne, “The super-excess energy dissipation in diatom algae: comparative analysis with higher plants,” Photosynth. Res. 82(2), 165–175 (2004). [CrossRef]
46. A. Chekalyuk and M. Hafez, “Photo-physiological variability in phytoplankton chlorophyll fluorescence and assessment of chlorophyll concentration,” Opt. Express 19(23), 22643–22658 (2011). [CrossRef]
47. P. Coppo, A. Taiti, L. Pettinato, M. Francois, M. Taccola, and M. Drusch, “Fluorescence imaging spectrometer (FLORIS) for ESA FLEX mission,” Remote Sens. 9(7), 649 (2017). [CrossRef]
48. D. Bouffard, D. Odermatt, J. Runnalls, and J. Sukys, “Heterogeneous data platform for operational modeling and forecasting of Swiss lakes,” Datalakes (2022), https://www.datalakes-eawag.ch.
