White Sea

Satellite data

The whole basin was divided into 7 sub-regions according to position of main sources of suspended matter and processes forming TSM distribution (Fig. 8) (Burenkov et al. 2004). Fig.9 shows coverage of all of the sub-regions by SeaWiFS and MODIS-Aqua data.

As it was mentioned in Introduction, the time series of bio-optical characteristics of the White Sea have been extended by including SeaWiFS data since 1998. Unfortunately, the coverage by SeaWiFS data is much worse than from MODIS, especially for May.

Fig_9a    Fig_9b


Development of the regional algorithms is usually based on concurrent field measurements of normalized water-leaving radiances LNW and of concentration of seawater constituents (total suspended matter TSM and chlorophyll-a Chl concentrations). Our field studies were carried out in the White Sea in the period of 2004-2010. But for various reasons, primarily because of the weather conditions, there was no high-quality match-up in the White Sea. So a modified approach was applied to derive the regional algorithms in the White Sea for retrieval of Chl and TSM concentrations from satellite data. It was as follows.

All in situ data on
Chl and TSM in the near-surface layer for the period 2004-2010 were collected, including data from measurements during the ship’s movement, which accounted for a significant part of the total number of measurements. Each point with available values of Chl and TSM was considered as a potentially sub-satellite point, and for it the MODIS data were chosen. Only data, corresponding to good weather conditions, were used in further analysis (quality check of satellite data was carried out).

Chlorophyll concentration

Direct determination of concentration of phytoplankton pigments (chlorophyll “a” and pheophytin “a”) was performed by fluorometric method with seawater samples (Kravchishina et al. 2011). After a preliminary rejection, 68 pairs of data of simultaneous measurements of chlorophyll concentration and of the seawater radiance reflectance from MODIS-Aqua (after reprocessing of 2009 and 2010) were selected to develop a regional algorithm. The following regression equation was obtained:

Chl = 2.13 (RRS(531)/RRS(547))-2.42,                                   (5),

where RRS(531) and RRS(547) are values of the above-surface remote-sensing reflectance; n=68, r2=0.61.

For SeaWiFS data the equation transformed to

Chl = 1.9 (RRS(510)/RRS(555))-0.87.                                       (5a)

Particle backscattering and TSM

Similar approach was applied to derive TSM concentration. Its values were calculated in two steps: first, values of the particle backscattering coefficient bbp were retrieved from satellite data (see Section Barents Sea, Algorithms), second, the TSM concentration was derived by using a relationship between TSM and bbp.

To derive such a relationship for the White Sea, 195 pairs of quasi-concurrent measurements of
bbp and TSM were selected. After standard statistical calculation, the following regression equation was derived:

TSM=22.8 bbp 0.53 ,                                                        (6)

where TSM is measured in mg/l, bbp – in m-1; n=195, r2=0.70.

Yellow substance absorption

To derive values of the yellow substance absorption coefficient ag, the semi-analytic bio-optical algorithm was used (Burenkov et al. 2001b). Due to small values of the water-leaving radiance in the shortwave spectral region in the White Sea and great errors of the atmospheric correction algorithm there, the MODIS spectral bands 412, 443, 469, 488 nm were not used, and the input parameters for solving the inverse problem were values of the above-surface remote-sensing reflectance Rrs(λi) at six MODIS spectral bands 531, 547, 555, 645, 667, and 678 nm. The values of Rrs(λi) were converted into values of the subsurface radiance reflectance ρ(λi) by means of a formula given in Lee et al. (1998).

The inverse problem was solved by a least square method, and, as a result of its solving, three unknowns were retrieved: chlorophyll concentration Chl, the particle backscattering coefficient bbp. and the yellow substance absorption coefficient ag.
As to Chl and bbp, preference was given to the regression algorithms, described above, as more reliable. Validation of the semi-analytic algorithm for ag was performed by using in situ measured values of the diffuse attenuation coefficient Kd.

Results and discussion

The mean monthly distributions of Chl, the particle backscattering and yellow substance absorption coefficients, TSM and Sea Surface Temperature (SST), derived from MODIS-Aqua data, are presented on color maps. Analysis of these maps shows that the areas of highest values of the bio-optical parameters (Chl, bbp, ag, and TSM) are associated with the North Dvina, Onega, Mezen Rivers runoff. As to Chl, it can be explained by nutrient supply from river runoff.

With respect to TSM, it can be seen from the maps, that the areas of turbid waters near the deltas of Onega and Mezen are larger than near North Dvina, although the discharge of North Dvina is greater than of two others. This can be explained by intensive tidal currents in the Onega and Mezen Bays, responsible for seashore abrasion and re-suspension of bottom sediments. It is well seen that turbid North Dvina waters propagate mainly along the eastern White Sea coast and TSM decreases rapidly towards the central part of the Dvina Bay. This is explained by along-shore cyclonic circulation in the White Sea, preventing propagation of turbid waters towards the central part of the sea. Only in May when rivers runoff is most intensive, North Dvina waters partly propagate towards the central part of the White Sea. Due to cyclonic circulation relatively transparent Barents Sea waters propagate along the western White Sea coast. The lowest values of TSM are observed in the central part of the sea, in the Kandalaksha Bay (where the river runoff is small). Values of TSM in these areas in August-September are less than 0.5-0.8 mg/l. Note that in the north-western part of the White Sea near the boundary with the Barents Sea high values of TSM are observed in September and most pronounced in 2003, 2008, 2010, 2012. This is associated with coccolithophore bloom in the Barents Sea (see Barents Sea, Result and discussion) which waters partly propagate into the White Sea.

Spatial variability of chlorophyll concentration is usually weaker than of TSM. As a whole spatial distributions of Chl are quite similar to distributions of TSM. The highest values of Chl (usually more than 2 mg/m3) are observed in Dvina, Onega and Mezen Bays and the lowest – in central part of the sea.

The seasonal variability of suspended matter concentration is well pronounced in the central part of the sea, in Kandalaksha and Dvina Bays. Maximum values of 
TSM are observed in May (sometimes in June) during spring flood-time when river waters occupy the most part of Dvina Bay and partly propagate into the central parts of the sea. With decrease of river discharge the area of turbid waters in the Dvina Bay is reduced, and they propagate mainly along the eastern coast of the White Sea. This leads to decrease of TSM in the open parts of the sea and in Kandalaksha Bay.

This qualitative reasoning is confirmed by quantitative estimation of suspended matter content in different regions of the White Sea. Seasonal changes of 
TSM, as well as of Chlbbpag, and SST in different sub-regions during the observational period (2002-2010) are shown in Fig.10, 11, 12, 13, 14.

Fig_10a Fig_10b   Fig_11a Fig_11b
Fig_12a Fig_12b
Fig_13a Fig_13b  Fig_14a Fig_14b

The most intensive seasonal changes (about 50%) are observed in the Dvina Bay in accordance with above reasoning. Maximal values of TSM here take place in spring and sometimes in autumn, minimal – in summer. In the central part of the White Sea and in Kandalaksha Bay increased values of TSM are observed only in May when river waters penetrate to the open sea. In other sub-regions seasonal changes are less. In addition to seasonal changes of river runoff they are influenced by other processes mentioned above.

In respect high values of ag, the White Sea is much superior to other Russian seas, even if they are also under strong influence of the river runoff. It is interesting to note that in the region of the Danube shelf in the Black Sea and in the Northern Caspian, which is under strong influence of Volga river run-off, the yellow absorption coefficients are of the same order – the monthly means do not exceed 0.3 m-1, whereas in the White Sea the ag values are more than 0.5 m-1, even in the open part, and reach the values of greater than 3 m-1 in the Dvina Bay. This is obviously connected with the characteristics of the catchment area of Dvina and other rivers flowing into the White Sea.

Such large values of the absorption coefficient prevent the penetration of solar radiation into the water column and it is a reason of low values of the radiance reflectance in the White Sea.

The seasonal (May-September) mean values with their standard deviations for Chl, the particle backscattering (bbp) and the yellow substance absorption (ag) coefficients, TSM, in different regions of the White Sea in 1998-2012, and SST in 2003-2012 are given in Table 2. Spatial and temporal variability (seasonal and inter-annual) of chlorophyll and TSM concentrations in the White Sea from satellite MODIS-Aqua data is considered in Burenkov et al (2011b).