Light Absorption by Optically Active Components in the Arctic Region (August 2020) and the Possibility of Application to Satellite Products for Water Quality Assessment

Abstract

In August 2020, during the 80th cruise of the R/V “Akademik Mstislav Keldysh”, the chlorophyll a concentration (Chl-a) and spectral coefficients of light absorption by phytoplankton pigments, non-algal particles (NAP) and colored dissolved organic matter (CDOM) were measured in the Norwegian Sea, the Barents Sea and the adjacent area of the Arctic Ocean. It was shown that the spatial distribution of the three light-absorbing components in the explored Arctic region was non-homogenous. It was revealed that CDOM contributed largely to the total non-water light absorption (a_tot(λ) = a_ph(λ) + a_NAP(λ) + a_CDOM(λ)) in the blue spectral range in the Arctic Ocean and the Barents Sea. The fraction of NAP in the total non-water absorption was low (less than 20%). The depth of the euphotic zone depended on a_tot(λ) in the surface water layer, which was described by a power equation. The Arctic Ocean, the Norwegian Sea and the Barents Sea did not differ in the Chl-a-specific light absorption coefficients of phytoplankton. In the blue maximum of phytoplankton absorption spectra, Chl-a-specific light absorption coefficients of phytoplankton in the upper mixed layer (UML) were higher than those below the UML. Relationships between phytoplankton absorption coefficients and Chl-a were derived by least squares fitting to power functions for the whole visible domain with a 1 nm interval. The OCI, OC3 and GIOP algorithms were validated using a database of co-located results (day-to-day) of in situ measurements (n = 63) and the ocean color scanner data: the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Terra (EOS AM) and Aqua (EOS PM) satellites, the Visible and Infrared Imager/Radiometer Suite (VIIRS) onboard the Suomi National Polar-orbiting Partnership (S-NPP) and JPSS-1 satellites (also known as NOAA-20), and the Ocean and the Land Color Imager (OLCI) onboard the Sentinel-3A and Sentinel-3B satellites. The comparison showed that despite the technological progress in optical scanners and the algorithms refinement, the considered standard products (chlor_a, chl_ocx, aph_443, adg_443) carried little information about inherent optical properties in Arctic waters. Based on the statistic metrics (Bias, MdAD, MAE and RMSE), it was concluded that refinement of the algorithm for retrieval of water bio-optical properties based on remote sensing data was required for the Arctic region.

1. Introduction

At present, the global ocean, covering about 70 percent of the Earth’s surface, is affected by climate change [1], which can significantly affect the physical and biological parameters of the ocean. Numerous recent studies have reported that the Arctic region was warming two-to-three times as fast as the global average [2,3,4]. According to present data, for the last 43 years (since 1979), the Arctic has been warming nearly four times faster than the globe [5]. Since 1999, the Arctic Ocean has experienced rapid sea ice loss [6]. The permanent loss of multiyear sea ice has resulted in an increase of sea surface temperature [7] and cloudiness [8], which has caused a decrease in photosynthetically active radiation (PAR) incident on the sea surface [9]. Moreover, changes in nutrient cycling [10], in phytoplankton composition [11] and in net primary production [12] occurred due to global warming in the Arctic region. Climate change is a serious threat to Arctic biodiversity: the potential of introducing nonindigenous species to Arctic seas is increasing, but many native polar species may not be able to tolerate warming [13]. In this regard, operative monitoring and understanding of the current state of the water ecosystem in the Arctic region is necessary.

Phytoplankton are the base of the food webs of the oceanic ecosystems. The rate of primary production of phytoplankton controls the chemical energy flow to higher trophic levels [14]. Water productivity indicators such as chlorophyll a and primary production can be assessed by ocean color remote sensing from satellites. Remote assessment of the primary production is used widely [15,16,17] due to unique opportunities of the remote approach, which provides a great spatial and temporal coverage. However, standard ocean color algorithms do not work well in the Arctic waters [18,19,20] due to a number of difficulties and intrinsic limitations, including higher concentrations of colored dissolved organic matter (CDOM) caused mostly by river runoff [21,22,23] and lower chlorophyll-a-specific absorption coefficients [24,25]. CDOM and non-algal particles (NAP) are poorly correlated to chlorophyll a in the Arctic [21], justifying that Arctic waters can be optically complex and the assumption for Case 1 waters are often not met [15].

Because of this, assessment of water productivity based on satellite data requires the development of regional Arctic algorithms based on regional peculiarities in inherent optical property (IOP) values and large solar zenith angles [16,26,27]. Thence, the IOPs, in particular, light absorption by optically significant substances (phytoplankton, NAP and CDOM [28]) are required to be measured and analyzed for the retrieval of empirical regularities of spatial and temporal variability of the spectral light absorption coefficient of the phytoplankton, NAP and CDOM. These regularities are necessary to assess the primary production by a full spectral approach [29] which takes into account the effect of spectral characteristics of both downwelling irradiance and phytoplankton absorption coefficients (a_ph(λ)) on the primary production.

Up to now, bio-optical data for the Arctic Ocean are not well documented [15]. In recent times, the IOPs of the western Arctic Ocean were analyzed in detail by Matsuoka et al. [30]. It was shown that values of the a_ph(λ) and the NAP absorption coefficient (a_NAP(λ)) at a wavelength of 440 nm varied widely from 0.001 to 0.2 m⁻¹ [30]. It was found [30] that values of the CDOM absorption coefficient (a_CDOM(440)) (0.005 to 0.5 m⁻¹) exceeded the a_ph(λ) and a_NAP(λ) values. The a_CDOM(440) observed in the Arctic region fell within the range obtained for coastal waters around Europe [28], with the exception of the lowest values, which were consistent with open oceanic waters [31]. Other published data, which were obtained in the Arctic region up to 2011 [22,32,33,34,35,36], confirmed the results of the study by Matsuoka et al. (2011) [30].

However, the bio-optical properties of Arctic waters are rapidly changing due to climatic effects. Sea ice melting releases an appreciable quantity of organic particles [37]. High loads of organic and inorganic dissolved and particulate matter are introduced in Arctic waters with river runoff through the Siberian continental shelves [38,39,40]. Changes in the composition and concentration of particulate and dissolved matter affect the propagation of spectral downwelling irradiance through the water column [26]. NAP absorption is strongest in the blue spectrum range, decreasing exponentially to the red spectrum range. CDOM absorption is similar to that of NAP due to the similarity in composition (organic matter) but exhibits a steeper exponential slope [26]. It was shown that in the Beaufort Sea, the particle enrichment flattened a remote sensing reflectance (Rrs) spectrum by reducing the Rrs by up to 20% in the spectral range from 400 to 550 nm [37]. Due to the dependence of apparent optical properties, including spectral Rrs, on the IOPs [26], the assessment of the current state and spatial variability of the IOPs in Arctic waters is especially relevant for remote sensing algorithm development [41].

The aim of this research was to assess the spatial distribution of spectral light absorption coefficients by optically active components (phytoplankton, NAP and CDOM) and their contribution to total non-water absorption in the Norwegian Sea, Barents Sea and the Arctic Ocean in August 2020.

2. Materials and Methods

2.1. Water Sampling

Bio-optical data were collected in August 2020 in the Barents Sea, the Norwegian Sea and the Arctic Ocean during the 80th cruise of the R/V “Akademik Mstislav Keldysh” [42] (Figure 1). Water sampling was carried out during daylight hours. Water samples were collected at 5 to 7 depths of the euphotic zone (Z_eu), depending on the hydrological water structure. Water sampling and CTD sounding were done with SBE oceanographic complex. Z_eu was defined as the depth of the 1% surface PAR level. PAR was measured with the LI-COR device, consisting of photodiode sensors LI-192 (for measuring underwater irradiance) and LI-190SA (for monitoring the level of PAR incident on the sea surface). Also, water transparency was assessed by Secchi disk depth (Zs).

The upper mixed layer (UML) depth was calculated as the depth at which potential density was different from the sea surface density by 0.05 kgm⁻³.

PAR in the UML (PAR_UML) was calculated in accordance with [43]:

PAR_UML = PAR₀ × (1 − e^{(−4.6 Z}_UML^/Z_eu⁾)/(4.6 Z_UML/Z_eu),

where PAR₀ is daily PAR incident on the sea surface and Z_UML is the depth of the upper mixed layer.

2.2. Pigment Analysis

The sum of chlorophyll a and phaeopigment concentrations (Chl-a) was determined by the spectrophotometric method [44,45] using a dual-beam spectrophotometer LAMBDA 35 (PerkinElmer, Waltham, MA, USA). Water samples (1–2 L) were filtered through Whatman GF/F filters (filter diameter −25 mm) with a pore diameter of 0.7 μm under low vacuum (<25 kPa). After the filtration, filters were wrapped in aluminum foil, flash-frozen in liquid nitrogen and kept at −80 °C until analysis in a laboratory. In the laboratory, filters were placed in 5 mL of 90% acetone in a glass centrifuge tube, then treated with vibration for 20 s using a vibration mixer (FALK instruments, Treviglio, Italy), kept at 5 °C for 10 h and then centrifuged. For a more complete extraction of phytoplankton pigments, the procedure was repeated.

2.3. Absorption Measurements

The particulate absorption coefficients were measured by the “quantitative techniques on wet filters” [24] in accordance with [46]. A quantity of 1–2 L of seawater was filtered through glass-fiber filters (Whatman GF/F) (filter diameter 25 mm) with a pore diameter of 0.7 μm under low vacuum (<25 kPa) shortly after sampling (<2 h). The spectral optical density (OD_p(λ)) of the particles on the filter was measured using a dual-beam spectrophotometer, LAMBDA 35 (PerkinElmer), equipped with a Spectralon integrating sphere, from 400–750 nm in 1 nm increments. The OD_p(λ) was converted to the spectral particulate absorption coefficient (a_p(λ)):

a_p(λ) = 2.303 × OD_p(λ)/(V/S),

where 2.303 = ln10, V is the water filtration volume (in m³), and S is the filter clearance area (in m²).

Then phytoplankton pigments were extracted in hot methanol [47], and the OD_NAP(λ) of non-algal particles was measured and converted to a_NAP(λ):

a_NAP(λ) = 2.303 × OD_NAP(λ)/(V/S).

The absorption coefficient of phytoplankton (a_ph(λ)) was obtained by subtracting a_NAP(λ) from a_p(λ). The path length amplification factor (β-correction) was estimated applying the quadratic equation described in [48]. To get the Chl-a-specific light absorption coefficient of phytoplankton ($a_{ph}^{*}\left(\mathsf{\lambda }\right)$) the values of a_ph(λ) were divided by Chl-a. Relationships between a_ph(λ) and Chl-a were derived by least squares fitting to power functions for the visible spectral domain (from 400 to 700 nm) in 1 nm increments. Spectral distributions of a_NAP(λ) coefficients were described with an exponential function [28]. Variations in the spectral shape of NAP absorption were described using the spectral slope (S_NAP). The S_NAP was calculated by fitting data to the exponential function for the visible spectral domain (from 400 to 700 nm).

CDOM light absorption coefficients (a_CDOM(λ)) were measured in accordance with modern NASA protocol [46]. Water samples were filtered through a nucleopore filter (Sartorius) with a pore diameter of 0.2 μm, using GF/F filters for prefiltration. The filters were pre-washed with ~50 mL of deionized water. The sample OD(λ) was measured opposite deionized water using quartz cuvettes with an optical path length of 0.1 m. The OD(λ) was also measured from 250 to 750 nm in 1 nm increments using a dual-beam spectrophotometer, LAMBDA 35 (Perkin Elmer). Spectral distributions of a_CDOM(λ) were described by an exponential function [28]. Slope coefficient (S_CDOM) was calculated by fitting data to the exponential function for the wavelength range from 350 to 500 nm.

2.4. Satellite Data

We used Level-2 Moderate Resolution Imaging Spectroradiometer (MODIS), Visible and Infrared Imager/Radiometer Suite (VIIRS) and Ocean and Land Color Imager (OLCI) satellite data from OceanColor Web (https://oceancolor.gsfc.nasa.gov/, accessed on 22 June 2021 (MODIS and VIIRS) and 20 April 2023 (OLCI)). The satellite data were chosen day-to-day with in situ measurements.

chlor_a (mg m⁻³)—chlorophyll concentration, OCI algorithm [49].

chl_ocx (mg m⁻³)—chlorophyll concentration, OC3 algorithm [50].

aph_443 (m⁻¹)—light absorption by phytoplankton at 443 nm, calculated using the default global configuration of the Generalized Inherent Optical Property (GIOP) model [51,52].

adg_443 (m⁻¹)—light absorption by NAP plus CDOM at 443 nm, calculated using GIOP model [51,52].

2.5. Statistical Analysis

Mapping and graph plotting were performed using QGIS desktop 3.8.0 and Grapher v.11 software, respectively. Statistical analyses were performed using the Microsoft Excel software package. The correlation analysis was performed with confidence probability of 0.05. To determine whether there was a statistically significant difference between the means in two unrelated data groups, the independent samples t-test with the significance level α = 0.05 was used. The criterion of signs (p-value) of less than 0.05 was considered statistically significant.

Systemic bias was calculated as:

$$\mathrm{b}\mathrm{i}\mathrm{a}\mathrm{s}=10^\left(\frac{{\sum }_{i=1}^n{\mathit{log}}_{10}\left({Sat}_i\right)-{\mathit{log}}_{10}\left({InSitu}_i\right)}{n}\right).$$

Median Absolute Difference was calculated as:

$$\mathrm{M}\mathrm{d}\mathrm{A}\mathrm{D}={10}^{Median\left|{\log }_{10}\left({Sat}_i\right)-{\log }_{10}\left({InSitu}_i\right)\right|}$$

Mean Absolute Error was calculated as:

$$\mathrm{M}\mathrm{A}\mathrm{E}=10^\left(\frac{{\sum }_{i=1}^n\left|{log}_{10}\left({Sat}_i\right)-{log}_{10}\left({InSitu}_i\right)\right|}{n}\right)$$

The parameter of a linear regression fitting, Root Mean Square Error, was calculated as:

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\sqrt{\frac{{\sum }_{i=1}^n{\left({Sat}_i-{InSitu}_i\right)}^2}{n}}$$

where n is the number of samples, Sat_i is the satellite data and InSitu_i is the in situ measured data. The best result is considered when bias, MdAD and MAE are equal to 1 and RMSE is equal to 0.

The statistical parameters for the products Chl-a and colored detrital matter (CDM) were computed following a log-normal hypothesis.

3. Results

3.1. Pigment Concentration

In the Arctic Ocean (16–19 August 2020), surface water temperature varied from −1.4 to 3.3 °C. The Chl-a values in the UML (Figure 2) varied between stations over a wide range (0.0066–1.3 mg m⁻³) and were, on average, 0.43 ± 0.54 mg m⁻³. The UML was in a range from 2 to 20 with mean equal to 10 ± 6 m. The maximum temperature gradient in the Arctic Ocean varied from 0.080 to 1.0 °C m⁻¹ and was, on average, 0.56 °C m⁻¹. The temperature gradient was located at 3.5–21 m (at 13 ± 6 m on average). Z_eu values varied from 29 to 60 m with mean equal to 43 ± 10 m. At some stations, the deep Chl-a maximum was observed in the thermocline, located within the euphotic zone (Figure 2).

During the first leg of the expedition (16–19 August), water temperatures in the surface layer of the Barents Sea were lower (3.7–5.7 °C) than temperatures on 21–23 August (5.7–11 °C). The UML was located at 6–24 m (mean depth of 15 ± 5 m), which was deeper than in the Arctic Ocean. The Chl-a values in the UML (Figure 2) changed within a narrower range (from 0.071 to 0.73 mg m⁻³ with mean 0.41 ± 0.26 mg m⁻³) than in the Arctic Ocean. In the Barents Sea, the seasonal thermocline was well developed, with a temperature gradient from 0.68 °C m⁻¹ to 1.5 °C m⁻¹ (on average 0.87 °C m⁻¹). The thermocline was located at 13–35 m (mean 23 ± 7 m) depth, while Z_eu values varied from 21 to 50 m (mean equal to 36 ± 11 m). Similarly to the Arctic Ocean, in the Barents Sea, the deep Chl-a maximum at some stations was located below the thermocline.

In the Norwegian Sea (5–13 August 2020), surface water temperature values were the highest (8.0–13 °C) among the study water areas. The UML was in a range 11–22 m, with 14 ± 4 m on average. The Chl-a values in the UML (0.54–2.0 mg m⁻³) were highest in comparison with the Barents Sea and the Arctic Ocean (Figure 2). The average Chl-a value (1.0 ± 0.44 mg m⁻³) in the Norwegian Sea was about twice as high as in both the Barents Sea and the Arctic Ocean. The maximum temperature gradient varied from 0.31 °C m⁻¹ to 1.3 °C m⁻¹ (on average, 0.74 °C m⁻¹) and was similar to that observed in the Barents Sea. The temperature gradient was located at 12–30 m (mean 19 ± 8 m). Water transparency in the Norwegian Sea was comparable with water transparency in the Arctic Ocean and the Barents Sea: Z_eu varied from 26 to 35 m and the mean was equal to 31 ± 4 m.

3.2. Light Absorption by Phytoplankton

Results were grouped into two datasets: (1) UML dataset and (2) below UML dataset within the euphotic zone, because seasonal water stratification divided the euphotic zone into two quasi-isolated layers. The variability of the spectral light absorption coefficients of phytoplankton, NAP and CDOM were analyzed for these two layers. The blue wavelength 438 nm was chosen for analysis because all three optically active components absorbed light significantly at the blue spectrum range and because this wavelength corresponded to the phytoplankton absorption maximum.

The a_ph(λ) spectrum shape was characterized by two main peaks: the blue maximum at a wavelength of ~438 nm and the red maximum at ~678 nm (Figure 3). In the UML, the a_ph(λ) values at wavelengths of 438 nm (a_ph(438)) and 678 nm (a_ph(678)) in the Norwegian Sea varied from 0.027 to 0.12 m⁻¹ and from 0.012 ± 0.051 m⁻¹, respectively. In the Norwegian Sea, a_ph(438) and a_ph(678) were higher than in both the Barents Sea and the Arctic Ocean. In the Barents Sea, the a_ph(438) and a_ph(678) values were in a range from 0.0030 to 0.049 m⁻¹ and from 0.0012 to 0.021 m⁻¹, respectively. In the Arctic Ocean, the a_ph(438) and a_ph(678) values varied from 0.0034 to 0.069 m⁻¹ and from 0.0016 to 0.041 m⁻¹, respectively. In the vertical profile of a_ph(λ), the highest values were associated with the deep Chl-a maximum located below UML (Figure 3).

Despite the marked difference in absorption coefficients between the seas, the shape of the spectrum, which was estimated by the ratio between absorption at the blue and red peaks (R_ph), was not different between the seas. The R_ph values decreased with depth from average 2.3 ± 0.35 in the UML to 2.0 ± 0.24 below the UML. This difference in the R_ph values between the UML and below UML was statistically significant (p < 0.00001).

The a_ph(λ) was strongly correlated to Chl-a. The relationship between the a_ph(438) and a_ph(678) coefficients and Chl-a is described by a power function:

a_ph(λ) = A(λ) × Chl-a ^B(λ),

(1)

where A(λ) is the spectral coefficient, equal to $a_{ph}^{*}\left(\mathsf{\lambda }\right)$) when Chl-a equal to 1 mg m⁻³; and B(λ) is the spectral power coefficient (in related unit).

The obtained A(λ) and B(λ) coefficients in the relationship linking the light absorption coefficients a_ph(438) and a_ph(678) with Chl-a (Equation (1)) turned out to be the same for the Arctic Ocean, the Norwegian Sea and the Barents Sea. But these coefficients differed between the water layers of the euphotic zone: UML and below UML (Figure 4).

To retrieve spectral distribution of a_ph(λ) coefficients from Chl-a, the relationship between these parameters needs to be determined for the visible domain of radiance (from 400 to 700 nm). The a_ph(λ) vs. Chl-a dependencies for the UML and below UML datasets were parameterized with 1 nm spectral resolution using Equation (1) for the Arctic region in summer. The results of the parameterization are presented in Figure 5 and in

Table 1. Spectral values of the constants A(λ) and B(λ) obtained when fitting the variations of a_ph(λ) vs. the sum of chlorophyll a and phaeopigment concentration (Chl-a) to power laws of the form a_ph(λ) = A(λ) × (Chl-a)^B(λ) in the UML of the Arctic region in summer.

λ	A(λ)	B(λ)	λ	A(λ)	B(λ)	λ	A(λ)	B(λ)	λ	A(λ)	B(λ)
400	0.0367	0.7747	476	0.0438	0.9775	551	0.0076	0.8966	626	0.0071	0.8567
401	0.0371	0.7780	477	0.0431	0.9819	552	0.0073	0.8947	627	0.0072	0.8585
402	0.0376	0.7868	478	0.0424	0.9839	553	0.0071	0.8853	628	0.0073	0.8547
403	0.0381	0.7870	479	0.0415	0.9858	554	0.0069	0.8797	629	0.0074	0.8627
404	0.0388	0.7940	480	0.0408	0.9888	555	0.0067	0.8826	630	0.0075	0.8632
405	0.0393	0.7963	481	0.0401	0.9912	556	0.0065	0.8801	631	0.0076	0.8660
406	0.0400	0.8015	482	0.0394	0.9954	557	0.0063	0.8732	632	0.0077	0.8656
407	0.0406	0.8089	483	0.0386	0.9954	558	0.0062	0.8647	633	0.0078	0.8718
408	0.0414	0.8092	484	0.0379	0.9956	559	0.0060	0.8642	634	0.0079	0.8710
409	0.0421	0.8119	485	0.0371	0.9980	560	0.0059	0.8559	635	0.0080	0.8731
410	0.0428	0.8133	486	0.0363	0.9995	561	0.0057	0.8523	636	0.0081	0.8775
411	0.0434	0.8172	487	0.0356	0.9992	562	0.0056	0.8472	637	0.0082	0.8810
412	0.0438	0.8207	488	0.0348	0.9981	563	0.0055	0.8503	638	0.0083	0.8869
413	0.0445	0.8207	489	0.0340	0.9978	564	0.0054	0.8543	639	0.0083	0.8876
414	0.0449	0.8232	490	0.0333	0.9981	565	0.0054	0.8436	640	0.0084	0.8948
415	0.0453	0.8267	491	0.0325	0.9961	566	0.0053	0.8409	641	0.0084	0.8971
416	0.0458	0.8269	492	0.0317	0.9956	567	0.0053	0.8455	642	0.0084	0.9046
417	0.0460	0.8279	493	0.0309	0.9934	568	0.0053	0.8466	643	0.0084	0.9000
418	0.0464	0.8287	494	0.0302	0.9918	569	0.0053	0.8459	644	0.0084	0.9043
419	0.0466	0.8323	495	0.0294	0.9897	570	0.0053	0.8482	645	0.0084	0.9105
420	0.0469	0.8320	496	0.0286	0.9844	571	0.0053	0.8459	646	0.0084	0.9133
421	0.0472	0.8328	497	0.0279	0.9845	572	0.0053	0.8500	647	0.0084	0.9097
422	0.0471	0.8271	498	0.0272	0.9823	573	0.0054	0.8485	648	0.0084	0.9082
423	0.0476	0.8329	499	0.0265	0.9818	574	0.0054	0.8499	649	0.0084	0.9040
424	0.0479	0.8412	500	0.0258	0.9776	575	0.0055	0.8530	650	0.0085	0.8998
425	0.0484	0.8402	501	0.0251	0.9776	576	0.0056	0.8544	651	0.0085	0.8943
426	0.0488	0.8388	502	0.0245	0.9769	577	0.0056	0.8584	652	0.0087	0.8917
427	0.0495	0.8435	503	0.0239	0.9755	578	0.0057	0.8525	653	0.0088	0.8846
428	0.0500	0.8442	504	0.0233	0.9716	579	0.0058	0.8576	654	0.0090	0.8799
429	0.0507	0.8425	505	0.0227	0.9724	580	0.0059	0.8662	655	0.0093	0.8648
430	0.0513	0.8438	506	0.0222	0.9685	581	0.0060	0.8667	656	0.0097	0.8572
431	0.0521	0.8480	507	0.0216	0.9711	582	0.0061	0.8673	657	0.0102	0.8560
432	0.0528	0.8531	508	0.0212	0.9707	583	0.0062	0.8735	658	0.0107	0.8520
433	0.0535	0.8528	509	0.0207	0.9700	584	0.0063	0.8772	659	0.0114	0.8436
434	0.0542	0.8502	510	0.0202	0.9691	585	0.0064	0.8840	660	0.0122	0.8432
435	0.0546	0.8578	511	0.0198	0.9699	586	0.0065	0.8843	661	0.0130	0.8420
436	0.0550	0.8545	512	0.0193	0.9728	587	0.0066	0.8938	662	0.0140	0.8446
437	0.0553	0.8598	513	0.0189	0.9744	588	0.0066	0.8961	663	0.0151	0.8465
438	0.0555	0.8574	514	0.0185	0.9759	589	0.0067	0.9013	664	0.0162	0.8506
439	0.0555	0.8620	515	0.0181	0.9772	590	0.0067	0.9032	665	0.0174	0.8577
440	0.0552	0.8653	516	0.0177	0.9732	591	0.0067	0.9058	666	0.0185	0.8638
441	0.0551	0.8650	517	0.0173	0.9774	592	0.0067	0.9077	667	0.0197	0.8701
442	0.0546	0.8698	518	0.0170	0.9765	593	0.0067	0.9102	668	0.0208	0.8719
443	0.0541	0.8725	519	0.0166	0.9785	594	0.0067	0.9128	669	0.0218	0.8825
444	0.0537	0.8730	520	0.0163	0.9782	595	0.0066	0.9138	670	0.0228	0.8832
445	0.0530	0.8775	521	0.0159	0.9765	596	0.0065	0.9079	671	0.0235	0.8887
446	0.0525	0.8846	522	0.0156	0.9752	597	0.0064	0.9076	672	0.0241	0.8919
447	0.0519	0.8864	523	0.0153	0.9769	598	0.0064	0.9110	673	0.0245	0.8962
448	0.0516	0.8941	524	0.0149	0.9766	599	0.0063	0.9049	674	0.0246	0.9013
449	0.0511	0.8959	525	0.0146	0.9789	600	0.0062	0.9035	675	0.0245	0.8996
450	0.0507	0.9013	526	0.0143	0.9765	601	0.0061	0.9048	676	0.0242	0.9006
451	0.0504	0.9027	527	0.0140	0.9760	602	0.0061	0.8983	677	0.0237	0.9005
452	0.0502	0.9108	528	0.0137	0.9732	603	0.0060	0.8945	678	0.0229	0.9029
453	0.0501	0.9129	529	0.0133	0.9733	604	0.0060	0.8897	679	0.0220	0.9012
454	0.0499	0.9121	530	0.0131	0.9678	605	0.0059	0.8851	680	0.0207	0.9007
455	0.0499	0.9204	531	0.0127	0.9635	606	0.0059	0.8809	681	0.0194	0.8911
456	0.0500	0.9228	532	0.0124	0.9595	607	0.0059	0.8815	682	0.0179	0.8855
457	0.0499	0.9223	533	0.0121	0.9621	608	0.0060	0.8784	683	0.0163	0.8736
458	0.0500	0.9254	534	0.0118	0.9572	609	0.0060	0.8784	684	0.0147	0.8636
459	0.0500	0.9283	535	0.0115	0.9564	610	0.0060	0.8793	685	0.0131	0.8527
460	0.0500	0.9320	536	0.0113	0.9524	611	0.0061	0.8673	686	0.0116	0.8395
461	0.0500	0.9335	537	0.0110	0.9481	612	0.0062	0.8686	687	0.0102	0.8256
462	0.0498	0.9378	538	0.0107	0.9470	613	0.0063	0.8636	688	0.0089	0.8169
463	0.0498	0.9384	539	0.0105	0.9399	614	0.0063	0.8591	689	0.0078	0.8001
464	0.0497	0.9392	540	0.0102	0.9353	615	0.0064	0.8621	690	0.0067	0.7887
465	0.0495	0.9431	541	0.0099	0.9324	616	0.0065	0.8588	691	0.0058	0.7776
466	0.0492	0.9444	542	0.0097	0.9297	617	0.0065	0.8599	692	0.0050	0.7628
467	0.0489	0.9464	543	0.0094	0.9289	618	0.0066	0.8585	693	0.0043	0.7521
468	0.0485	0.9512	544	0.0092	0.9258	619	0.0067	0.8611	694	0.0038	0.7460
469	0.0481	0.9522	545	0.0090	0.9134	620	0.0068	0.8602	695	0.0033	0.7305
470	0.0476	0.9590	546	0.0087	0.9146	621	0.0068	0.8579	696	0.0029	0.7268
471	0.0470	0.9585	547	0.0085	0.9094	622	0.0069	0.8552	697	0.0026	0.7201
472	0.0465	0.9634	548	0.0083	0.9068	623	0.0069	0.8584	698	0.0023	0.7087
473	0.0458	0.9665	549	0.0080	0.9003	624	0.0070	0.8563	699	0.0020	0.6984
474	0.0452	0.9703	550	0.0078	0.8976	625	0.0071	0.8534	700	0.0018	0.6933
475	0.0445	0.9739

and

Table 2. Spectral values of the constants A(λ) and B(λ) obtained when fitting the variations of a_ph(λ) vs. the sum of chlorophyll a and phaeopigment concentration (Chl-a) to power laws of the form a_ph(λ) = A(λ) × (Chl-a)^B(λ) below UML (within euphotic zone) of the Arctic region in summer.

λ	A(λ)	B(λ)	λ	A(λ)	B(λ)	λ	A(λ)	B(λ)	λ	A(λ)	B(λ)
400	0.0311	0.9975	476	0.0374	0.8571	551	0.0085	0.9651	626	0.0071	1.0237
401	0.0315	0.9956	477	0.0369	0.8568	552	0.0082	0.9707	627	0.0073	1.0166
402	0.0321	0.9877	478	0.0365	0.8503	553	0.0080	0.9799	628	0.0073	1.0143
403	0.0325	0.9881	479	0.0360	0.8481	554	0.0077	0.9857	629	0.0074	1.0114
404	0.0329	0.9853	480	0.0355	0.8440	555	0.0075	0.9894	630	0.0075	1.0078
405	0.0335	0.9805	481	0.0350	0.8414	556	0.0073	0.9966	631	0.0077	1.0011
406	0.0341	0.9755	482	0.0345	0.8367	557	0.0071	1.0016	632	0.0077	0.9992
407	0.0346	0.9737	483	0.0340	0.8348	558	0.0069	1.0079	633	0.0078	0.9933
408	0.0352	0.9702	484	0.0335	0.8312	559	0.0066	1.0208	634	0.0080	0.9868
409	0.0357	0.9677	485	0.0330	0.8283	560	0.0064	1.0277	635	0.0080	0.9870
410	0.0362	0.9632	486	0.0324	0.8271	561	0.0062	1.0370	636	0.0081	0.9791
411	0.0367	0.9595	487	0.0318	0.8242	562	0.0061	1.0440	637	0.0082	0.9741
412	0.0371	0.9578	488	0.0313	0.8216	563	0.0060	1.0493	638	0.0082	0.9697
413	0.0375	0.9541	489	0.0307	0.8211	564	0.0058	1.0588	639	0.0083	0.9628
414	0.0378	0.9529	490	0.0301	0.8197	565	0.0057	1.0618	640	0.0083	0.9575
415	0.0382	0.9480	491	0.0295	0.8184	566	0.0056	1.0672	641	0.0084	0.9497
416	0.0384	0.9486	492	0.0290	0.8193	567	0.0056	1.0722	642	0.0084	0.9458
417	0.0387	0.9455	493	0.0284	0.8193	568	0.0055	1.0775	643	0.0083	0.9421
418	0.0389	0.9437	494	0.0278	0.8203	569	0.0055	1.0832	644	0.0084	0.9368
419	0.0390	0.9414	495	0.0272	0.8202	570	0.0054	1.0857	645	0.0084	0.9344
420	0.0392	0.9404	496	0.0266	0.8214	571	0.0054	1.0871	646	0.0083	0.9327
421	0.0394	0.9343	497	0.0260	0.8229	572	0.0054	1.0901	647	0.0083	0.9305
422	0.0398	0.9353	498	0.0255	0.8250	573	0.0055	1.0877	648	0.0083	0.9313
423	0.0398	0.9375	499	0.0250	0.8240	574	0.0055	1.0903	649	0.0083	0.9334
424	0.0401	0.9327	500	0.0244	0.8269	575	0.0055	1.0929	650	0.0084	0.9354
425	0.0403	0.9321	501	0.0239	0.8284	576	0.0056	1.0930	651	0.0085	0.9387
426	0.0407	0.9277	502	0.0234	0.8303	577	0.0056	1.0924	652	0.0086	0.9424
427	0.0410	0.9273	503	0.0230	0.8310	578	0.0057	1.0891	653	0.0088	0.9470
428	0.0413	0.9263	504	0.0225	0.8324	579	0.0058	1.0858	654	0.0090	0.9539
429	0.0417	0.9264	505	0.0220	0.8349	580	0.0059	1.0850	655	0.0093	0.9575
430	0.0422	0.9234	506	0.0216	0.8360	581	0.0059	1.0842	656	0.0096	0.9611
431	0.0426	0.9214	507	0.0212	0.8372	582	0.0061	1.0810	657	0.0101	0.9635
432	0.0434	0.9141	508	0.0208	0.8383	583	0.0062	1.0776	658	0.0107	0.9635
433	0.0437	0.9155	509	0.0204	0.8411	584	0.0063	1.0727	659	0.0113	0.9639
434	0.0441	0.9131	510	0.0201	0.8424	585	0.0063	1.0714	660	0.0120	0.9619
435	0.0445	0.9080	511	0.0197	0.8426	586	0.0064	1.0673	661	0.0128	0.9563
436	0.0445	0.9150	512	0.0194	0.8425	587	0.0065	1.0628	662	0.0137	0.9513
437	0.0449	0.9088	513	0.0191	0.8451	588	0.0066	1.0605	663	0.0147	0.9435
438	0.0448	0.9109	514	0.0188	0.8455	589	0.0066	1.0577	664	0.0158	0.9362
439	0.0450	0.9070	515	0.0184	0.8469	590	0.0067	1.0537	665	0.0168	0.9283
440	0.0447	0.9078	516	0.0181	0.8476	591	0.0067	1.0512	666	0.0179	0.9219
441	0.0446	0.9080	517	0.0178	0.8494	592	0.0067	1.0491	667	0.0190	0.9122
442	0.0443	0.9068	518	0.0175	0.8509	593	0.0067	1.0467	668	0.0200	0.9054
443	0.0439	0.9084	519	0.0172	0.8519	594	0.0067	1.0476	669	0.0209	0.8975
444	0.0435	0.9097	520	0.0170	0.8521	595	0.0066	1.0453	670	0.0216	0.8920
445	0.0431	0.9112	521	0.0167	0.8542	596	0.0065	1.0474	671	0.0223	0.8878
446	0.0427	0.9094	522	0.0164	0.8550	597	0.0065	1.0467	672	0.0227	0.8840
447	0.0423	0.9104	523	0.0161	0.8584	598	0.0064	1.0486	673	0.0231	0.8785
448	0.0419	0.9114	524	0.0158	0.8602	599	0.0063	1.0499	674	0.0231	0.8784
449	0.0414	0.9147	525	0.0156	0.8610	600	0.0062	1.0501	675	0.0230	0.8753
450	0.0411	0.9151	526	0.0153	0.8641	601	0.0061	1.0533	676	0.0227	0.8743
451	0.0409	0.9150	527	0.0150	0.8663	602	0.0061	1.0526	677	0.0222	0.8740
452	0.0407	0.9154	528	0.0147	0.8685	603	0.0061	1.0541	678	0.0215	0.8751
453	0.0406	0.9157	529	0.0144	0.8718	604	0.0060	1.0545	679	0.0207	0.8757
454	0.0405	0.9173	530	0.0141	0.8739	605	0.0060	1.0545	680	0.0196	0.8810
455	0.0406	0.9129	531	0.0138	0.8790	606	0.0060	1.0533	681	0.0184	0.8838
456	0.0404	0.9172	532	0.0136	0.8802	607	0.0060	1.0530	682	0.0171	0.8921
457	0.0405	0.9132	533	0.0133	0.8839	608	0.0061	1.0536	683	0.0158	0.8991
458	0.0405	0.9124	534	0.0130	0.8891	609	0.0061	1.0533	684	0.0143	0.9083
459	0.0405	0.9136	535	0.0127	0.8926	610	0.0062	1.0497	685	0.0130	0.9178
460	0.0405	0.9117	536	0.0124	0.8967	611	0.0062	1.0471	686	0.0116	0.9315
461	0.0406	0.9079	537	0.0122	0.8996	612	0.0063	1.0459	687	0.0103	0.9424
462	0.0406	0.9073	538	0.0119	0.9041	613	0.0064	1.0443	688	0.0091	0.9572
463	0.0406	0.9039	539	0.0116	0.9096	614	0.0065	1.0408	689	0.0080	0.9731
464	0.0405	0.9025	540	0.0114	0.9128	615	0.0065	1.0388	690	0.0070	0.9898
465	0.0405	0.8966	541	0.0111	0.9178	616	0.0066	1.0398	691	0.0061	1.0093
466	0.0403	0.8964	542	0.0108	0.9224	617	0.0067	1.0364	692	0.0053	1.0270
467	0.0402	0.8915	543	0.0105	0.9277	618	0.0067	1.0337	693	0.0046	1.0506
468	0.0400	0.8886	544	0.0103	0.9301	619	0.0068	1.0345	694	0.0040	1.0750
469	0.0398	0.8845	545	0.0100	0.9367	620	0.0069	1.0287	695	0.0035	1.0980
470	0.0396	0.8821	546	0.0098	0.9422	621	0.0069	1.0302	696	0.0031	1.1203
471	0.0393	0.8772	547	0.0095	0.9451	622	0.0069	1.0289	697	0.0027	1.1443
472	0.0390	0.8728	548	0.0092	0.9505	623	0.0070	1.0299	698	0.0024	1.1767
473	0.0386	0.8696	549	0.0090	0.9524	624	0.0070	1.0254	699	0.0022	1.1987
474	0.0382	0.8664	550	0.0087	0.9594	625	0.0071	1.0226	700	0.0019	1.2246
475	0.0378	0.8622

.

3.3. Light Absorption by Non-Algal Particles

Light absorption by NAP in all investigated areas in the Arctic region co-varied with phytoplankton light absorption. In the UML, values of the a_NAP(438) in the Norwegian Sea varied from 0.011 to 0.039 m⁻¹, and were higher than in the Barents Sea (0.0036–0.032 m⁻¹) and the Arctic Ocean (0.0037–0.023 m⁻¹) (Figure 3). The differences in the NAP contribution to particulate light absorption between the water layers (UML and below UML) were not statistically significant (p = 0.22). At the wavelength of the blue maximum of phytoplankton absorption spectra, the contribution of a_NAP(438) to the a_p(438) was, on average, 29 ± 12%. In all investigated areas, spectral slope (S_NAP) varied in almost the same range (from 0.005 to 0.018 nm⁻¹) and was equal to 0.011 ± 0.003 nm⁻¹ on average.

3.4. Light Absorption by Colored Dissolved Organic Matter

The optical properties of the UML in the Arctic Ocean were characterized by lower CDOM absorption (a_CDOM(438)) values (0.013–0.063 m⁻¹) than in the Norwegian Sea (0.016–0.14 m⁻¹) and the Barents Sea (from 0.014 to 0.18 m⁻¹) (Figure 3). Below the UML, in the Arctic Ocean, the CDOM absorption coefficients (0.013–0.16 m⁻¹) were higher than in the UML (0.013–0.063 m⁻¹) (p = 0.025). In the Norwegian Sea and the Barents Sea, the values of a_CDOM(438) below the UML varied from 0.018 to 0.12 m⁻¹ and from 0.0085 to 0.20 m⁻¹, respectively (Figure 3). The a_CDOM(438) varied by more than an order of magnitude in all investigated areas with an exception of the the UML in the Arctic Ocean, where a_CDOM(438) changed less (about four times).

The S_CDOM values varied from 0.0085 to 0.034 nm⁻¹ (0.017 ± 0.005 nm⁻¹ on average) and did not differ statistically (p = 0.19) between the UML and below the UML.

In the Arctic Ocean, the Barents Sea and the Norwegian Sea, the CDOM absorption did not co-vary with Chl-a (Figure 6) and consequently, also did not co-vary with phytoplankton absorption.

An inverse relationship was revealed between the a_CDOM(438) and the S_CDOM. The relationship was described by the power function (y = A x^B). (Figure 7). Absorption by CDM at 438 nm (a_CDM(438) = a_CDOM(438) + a_NAP(438)) varied in the range 0.012–0.020 m⁻¹ (on average, 0.075 ± 0.046 m⁻¹). The CDM absorption slope coefficient (S_CDM) changed from 0.009 to 0.031 nm⁻¹ (on average 0.016 ± 0.0038 nm⁻¹). Between the a_CDM(438) and S_CDM, a relationship was revealed (Figure 7). Values of the power coefficient (B) of S_CDOM—a_CDOM(438) and S_CDM—a_CDM(438) relationships were almost the same (−0.34 and −0.32), but values of the A coefficient were different (0.0057 and 0.0062), which was related by a ratio between CDOM and CDM absorption coefficients.

3.5. Total Non-Water Light Absorption Budget

In the Arctic region, values of the total non-water light absorption coefficients by suspended and dissolved organic substances (a_tot(λ) = a_p(λ) + a_CDOM(λ)) were calculated for wavelengths of 438 nm (a_tot(438)) and 490 nm (a_tot(490)). These wavelengths were chosen because 438 nm corresponds to maximum of phytoplankton absorption and 490 nm corresponds to a channel used by almost all remote sensing optical scanners [55]. Values of the a_tot(438) varied from 0.021 to 0.25 m⁻¹ in the UML, and from 0.043 to 0.45 m⁻¹ in the layer below the UML. Values of the a_tot(490) were in range from 0.0086 to 0.16 m⁻¹ in the UML, and from 0.019 to 0.24 m⁻¹ below the UML. Absorption budgets for the UML and below UML layers are presented in Figure 8, using ternary plots that illustrate the relative contribution of each light-absorbing component to a_tot(438) and a_tot(490).

In the UML, the highest phytoplankton contribution to the total non-water absorption was obtained in the Norwegian Sea, bordering the Atlantic Ocean: on average, the a_ph(438) contribution was 45 ± 13% and that of a_ph(490) was 50 ± 15%. The lowest phytoplankton share in total absorption was detected in in the Arctic Ocean (25 ± 15% and 29 ± 19%, respectively) and in the Barents Sea (31 ± 12% and 33 ± 14%, respectively).

The average contributions of NAP to a_tot(438) and a_tot(490) in the UML of all investigated regions (in the Barents Sea, the Norwegian Sea and the Arctic Ocean) were 15 ± 7% and 18 ± 9%, respectively.

CDOM dominated the light absorption in the UML of the Arctic Ocean and the Barents Sea. The CDOM contributions at 438 and 490 nm were 61 ± 17% and 50 ± 20%, respectively, in the Arctic Ocean and 53 ± 17% and 49 ± 20%, respectively, in the Barents Sea. the contribution of CDOM to total light absorption was lowest in the Norwegian Sea: the a_CDOM(438) values were, on average, 40 ± 13%, and a_CDOM(490) values were, on average, 34 ± 14% (Figure 8).

The absorption budget in the layer below the UML differed from that of the UML in the phytoplankton and NAP contribution to total non-water light absorption. The phytoplankton dominated in the light absorption at 438 and 490 nm in the Arctic Ocean (42 ± 20% and 47 ± 21%, respectively) and in the Norwegian Sea (42 ± 17% and 50 ± 20%, respectively). The phytoplankton share reached ~60–70% in the deep Chl-a maximum in the Arctic Ocean. In the Barents Sea, phytoplankton contributed less at 438 and at 490 nm (29 ± 20% and 35 ± 24%, respectively) than in the Arctic Ocean and in the Norwegian Sea. But CDOM in the Barents Sea largely dominated in total non-water absorption at 438 nm 61 ± 24%) and at 490 nm (54 ± 27%). The NAP contribution to total absorption in the Barents Sea, the Norwegian Sea and the Arctic Ocean below the UML was generally low: the contributions at 438 and at 490 were 10 ± 5% and 11 ± 6%, respectively (Figure 8).

The variability in total non-water absorption resulted in a variation in water transparency by a factor of 2. An analysis of the dependence of water transparency on the water bio-optical properties revealed a relationship between Z_eu and total non-water light absorption in the surface layer. The relationship was described by a power equation (Figure 9).

3.6. Satellite Data

The ocean color scanner MODIS provided the most accurate retrievals of Chl-a (

Table 3. Validation of satellite retrieval Chl-a and light absorption coefficients: light absorption by phytoplankton pigments (aph_443) and colored detrital matter at 443 nm (adg_443). The bold font indicates the best results for each statistical metric.

Statistical Metric	MODIS				VIIRS				OLCI
	chlor_a	chl_ocx	aph_443	adg_443	chlor_a	chl_ocx	aph_443	adg_443	chlor_a	aph_443	adg_443
n	22	22	22	16	21	21	21	21	20	20	20
Bias	1.0	1.0	1.4	0.20	0.66	0.74	0.58	0.34	0.88	0.89	0.33
MAE	1.5	1.5	2.0	5.0	2.2	2.0	2.7	3.0	1.6	2.0	3.1
MdAD	1.3	1.3	1.5	5.7	2.0	2.1	2.4	2.6	1.5	1.8	2.4
RMSE	0.20	0.20	0.03	0.09	0.48	0.45	0.03	0.07	0.26	0.02	0.08

, Figure 10). There was practically no difference between chlor_a and chl_ocx: the RMSE was 0.2 mg m⁻³ and was equal in both directions (bias = 1.0). The least accurate Chl-a retrievals were obtained from the VIIRS scanner: the chlor_a and chl_ocx were significantly lower than the in situ Chl-a (bias varied from 0.66 to 0.74), and the RMSE varied from 0.45 to 0.48 mg m⁻³, with the average value of retrieved Chl-a equal to 0.34 mg m⁻³.

The phytoplankton light absorption coefficients at a wavelength of 443 nm were underestimated for the VIIRS and OLCI scanners (bias 0.58 and 0.89) and overestimated for MODIS (bias 1.4) (Table 3, Figure 10). For all color scanners, the RMSE ranged from 0.07 to 0.09 m⁻¹, with mean aph_443 values of 0.032, 0.016 and 0.021 m⁻¹ for MODIS, VIIRS and OLCI, respectively.

All algorithms tended to significantly underestimate CDM retrievals (Figure 10). Bias for adg_443 varied from 0.20 to 0.34, and RMSE varied from 0.02 to 0.03 m⁻¹ (Table 3).

4. Discussion

The investigations were carried out in the Barents Sea, the Norwegian Sea and the Arctic Ocean in summer 2020 in the post-bloom period of the year [56]. The obtained Chl-a values were typical for these Arctic regions in this season [56]. Within the UML, the Chl-a was distributed homogeneously. Spatial distribution of surface Chl-a was heterogeneous. The lowest Chl-a values were observed in the Arctic Ocean, and the highest Chl-a values were observed in the Norwegian Sea. Extremely low water temperatures (from −1.4 to 3.3 °C) were observed in the Arctic Ocean, and the highest water temperatures (from 8 to 13 °C) were observed in the Norwegian Sea. The spatial Chl-a co-varied with surface temperature. The temperature was likely to be crucial environment factor determining phytoplankton growth [57,58,59] and as result the Chl-a content in the Arctic region.

Phytoplankton absorption in all investigated area (the Arctic Ocean, the Barents Sea and the Norwegian Sea) in the UML and below UML co-varied with Chl-a. Despite the difference in Chl-a values between stations, the relation between Chl-a and a_ph(λ) was described by a power function with the same coefficients for the Arctic Ocean, the Norwegian Sea and the Barents Sea (Figure 4). Analysis of the link between phytoplankton light absorption coefficients and Chl-a revealed difference in chlorophyll-a-specific phytoplankton light absorption coefficient ($a_{ph}^{*}\left(\mathsf{\lambda }\right)$)) in the blue peak between layers: UML and below UML.

The A(438) coefficient in the power function (1) linking a_ph(438) with Chl-a in the UML and below UML were equal to 0.056 m² mg⁻¹ (Table 1) and 0.045 m² mg⁻¹ (Table 2), respectively. The difference between obtained coefficients was statistically significant (p = 0.0003). In the red peak of a_ph(λ) (at 678 nm), A(678) was equal to 0.024 m² mg⁻¹ in the UML (Table 1) and 0.023 m² mg⁻¹ below the UML (Table 2). The difference in A(678) between layers was insignificant (p = 0.09). The A(λ) values (Equation (1)) correspond to the $a_{ph}^{*}\left(\mathsf{\lambda }\right)$ when Chl-a was equal to 1 mg m⁻³. Thus, A(λ) changes reflect variability in $a_{ph}^{*}\left(\mathsf{\lambda }\right)$) that is highly variable due to phytoplankton acclimation to environmental factors [54,60]. In the Arctic region during the cruise (in August 2020), PAR₀ was, on average, 15 ± 4.7 E m⁻² day⁻¹. PAR profiles indicated that more than 80% of the PAR₀ attenuated within the UML. As a result, the PAR available for phytoplankton in the layer below the UML decreased. PAR_UML depends on PAR₀, water transparency and the ratio between Z_UML and Z_eu. PAR_UML was equal to 7.3 ± 2.3 E m⁻² day⁻¹. Below the UML, the maximum values of PAR were 2.8 ± 1.9 E m⁻² day⁻¹ and decreased to 0.15 ± 0.047 E m⁻² day⁻¹ with deepening to the bottom boundary of the Z_eu. Thus, the layers of the euphotic zone differed by almost an order of the PAR magnitude. The observed decrease in $a_{ph}^{*}\left(\mathsf{\lambda }\right)$ with depth was more pronounced in the blue peak of $a_{ph}^{*}\left(\mathsf{\lambda }\right)$ (at 438 nm) in comparison with the red peak (Figure 4). It resulted in R_ph decreasing with depth within the euphotic zone. It was shown that the variability of R_ph correlated with the intracellular relative (relatively to Chl-a) concentration of non-photosynthetic pigments absorbing light photons in the blue range [60].

The quota of photoprotective (non-photosynthetic) pigments increases due to algae photoacclimation to increasing PAR [61]. Moreover, it was found that, in the surface waters, an intracellular concentration of the photoprotective pigments tends to be greater at latitudes where PAR₀ is relatively high [62,63]. Consequently, the differences in A(438) values reflecting differences in $a_{ph}^{*}\left(438\right)$ are likely to be caused by phytoplankton acclimation (variation of pigment concentration and composition in the cells) to ambient light in the UML and below the UML [61,64]. The observed difference in A(438) and in $a_{ph}^{*}\left(438\right)$ between layers within the euphotic zone was in a good agreement with results obtained in the Black Sea in summer [53,65]. In the UML of the Black Sea, the A(440) (0.076 m² mg⁻¹) was about 1.6 times higher than that for the deep chlorophyll maximum (0.049 m² mg⁻¹) (Figure 5) [53,65].

The insignificant depth-dependent variability in$a_{ph}^{*}\left(678\right)$ observed in our research was likely related to the fact that the “pigment package effect”, associated with both intracellular pigment concentration and cell size (shift in phytoplankton species composition) [54,64,66,67,68,69], was invariable with depth. A decrease in the cell size of phytoplankton in the layer below the UML could compensate for the negative effect of an increase in the intracellular pigment concentration on $a_{ph}^{*}\left(438\right)$ [66]. Such an effect was observed in the Black Sea [53,65]. In general, the obtained $a_{ph}^{*}\left(\mathsf{\lambda }\right)$ values were within the range of previous reports for the Arctic region [33,37,70] and for global ocean [54].

The results showed a heterogeneity of the investigated water area in terms of spectral light absorption coefficients of optically active components and the ratio between them (Figure 3 and Figure 8). In general, the observed values of inherent optical properties of dissolved and suspended substances are in good agreement with the results obtained in the western Arctic region in summer by other researchers [30]. Total non-water absorption at 438 nm in the UML was dominated by CDOM in the Arctic Ocean and in the Barents Sea. Phytoplankton also contributed significantly and dominated in absorption in the Norwegian Sea.

The variability of the CDOM fraction in total non-water absorption was related to both the ice melting effect [37] and the phytoplankton abundance. Thus, in the most trophic Norwegian Sea, the average contribution of a_CDOM(438) (40%) was relatively less than those in the Arctic Ocean (56%) and the Barents Sea (55%). Nevertheless, our results confirm earlier observations in the Arctic region that CDOM is the dominant light-absorbing component almost everywhere [21,34,35] and not co-varying with Chl-a [37,71]. Based on all the results obtained in the Barents Sea, the Norwegian Sea and the Arctic Ocean, a link between the a_CDOM(438) (and a_CDM(438)) and S_CDOM (and S_CDM) values was revealed and described by a common exponential relationship (Figure 7). The relationship between a_CDOM(438) and the S_CDOM is in good agreement with observations in other regions of global oceans [72]. This negative relationship between a_CDOM(438) and S_CDOM is associated with a change in the relative composition of CDOM, namely, in the ratio between high-molecular-weight CDOM compounds and low-molecular-weight CDOM compounds, which resulted in the change in the S_CDOM [72]. The relationships between a_CDOM(438) (and a_CDM(438)) and the S_CDOM (and S_CDM) will allow retrieval of CDOM and CDM light absorption spectra based on the absorption coefficient at 438 nm. It could be used in development of a regional satellite algorithm using a three-band approach [73] successful in optically complex waters, as it was shown in the Black Sea examples [74].

The NAP contribution to total non-water absorption was generally low (less than 20%) (Figure 8). The low relative absorptions by NAP are typical for offshore regions of the world ocean which are not affected by coastal runoff [28,75]. The values of S_NAP slope were in good agreement with known data for different regions of the world ocean [28], including the Arctic region [30], which indicated a high degree of conservatism of S_NAP values.

A negative correlation between non-water total light absorption (phytoplankton, NAP and CDOM) in the surface layer and euphotic zone was revealed (Figure 9). The obtained relationship agreed with those for the Black Sea [76]. This relationship can be used for assessment of the euphotic zone based on the remote sensing data, if the three-band bio-optical algorithm [73] is applied for the Arctic region.

That parameterization of the light absorption by optically active components showed the high variability of CDM absorption and its domination in total non-water absorption. Consequently, Arctic waters are optically complex waters, where high uncertainty on Chl-a and IOP retrievals [16] is caused by the prevailing effect of the CDM on the remote sensing reflectance spectrum [77,78].

Due to the fact that the environment in the Arctic region is rapidly changing due to climatic effects [7,8,9,10,11,12], remote sensing data are required for operative tracking of changes in aquatic ecosystems. Using the obtained dataset, satellite data were compared with in situ data in order to assess the possibility of standard satellite products for assessment of water quality and productivity indicators in the Arctic waters. The comparison of coincident in situ measurements of Chl-a, a_CDM(443) and a_ph(443) with MODIS (Aqua and Terra), VIIRS (Suomi-NPP and NOAA-20 (JPSS-1)) and OLCI (Sentinel-3A and Sentinel-3B) satellite data showed slight agreement, indicating that the present algorithms carried little information about water quality and productivity indicators in the Arctic region (Figure 10, Table 3).

Correct assessment of the Chl-a in optically contrasting waters could be provided by the three-band algorithm [73], separating the light absorption by CDM and by phytoplankton following retrievals of Chl-a based on revealed link between a_ph(λ) and Chl-a [53]. The obtained parameterization of light absorption by optically active components can be used to adapt this three-band algorithm [73] for the Arctic waters to retrieve the bio-optical properties a_CDM(λ) and a_ph(λ). The relationship between the phytoplankton absorption coefficients and the Chl-a values revealed for the visible range with 1 nm increment (Table 1 and Table 2) can be used to retrieve Chl-a based on a_ph(λ). The a_ph(λ)-Chl-a parameterization can also be used in bio-optical models assessing downwelling irradiance and primary production using a full spectral approach [29,32].

5. Conclusions

In the Barents Sea, the Norwegian Sea and the Arctic Ocean, new data on the spatial distribution of chlorophyll a concentration (indicator of phytoplankton biomass and water productivity), spectral coefficients of light absorption by optically active components were obtained in summer (August 2020). Light absorption by optically active components was parameterized. The relationship between Chl-a and a_ph(λ) was revealed for the UML and the layer below the UML for the visible spectral domain (from 400 to 700 nm) with a 1 nm increment (Table 1 and Table 2). The Chl-a–a_ph(λ) relationship was described by a power function. The coefficients of this parameterization, in particular, the A(λ) coefficient, differed between layers in the euphotic zone due to phytoplankton photoacclimation (intracellular pigment concentration and composition). It should be noted that a_ph(λ) parametrization was revealed for a rather wide range of Chl-a: from 0.066 to 2.2 mg m⁻³ in the UML and from 0.14 to 4.6 mg m³ below the UML.

Values of Z_eu depended on total non-water light absorption in the surface layer, which was described by a power equation. The light absorption by phytoplankton was relatively high in the Norwegian Sea and the lowest in the Arctic Ocean. The colored dissolved organic matter mainly dominated in the total non-water absorption in the Arctic region, with the exception of the Norwegian Sea.

The OCI, OC3 and GIOP algorithms carried little information about Chl-a, a_CDM(λ) and a_ph(λ) in the Arctic waters. The parameterization of the light absorption by optically active components will allow the adaptation of the three-band algorithm developed for the retrieval of IOPs of the Black sea [73] to the Arctic. The revealed differences in the parametrization between two layers (UML dataset and below UML dataset) within the euphotic zone will provide more correct retrieval of the water productivity indicators. The development of such algorithms is relevant for prompt assessment of the Arctic ecosystem state under global climate change. The spectral approach to the assessment of phytoplankton photosynthesis rate will allow the assessment of the impact of ice melting on primary production, which determines the productivity of the pelagic ecosystem in general.