**3.1.
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.
7) [24].
Fig.8
shows coverage of all of the sub-regions by MODIS-Aqua data (see the
comment in 2.1).

Development of the regional
algorithms is usually based on concurrent field measurements of
normalized water-leaving radiances *L*_{WN} and of concentration of
seawater constituents (total suspended matter TSM and chlorophyll-a *Chl*
concentrations). Field measurements of *L*_{WN} in the White Sea are not numerous and carried out mainly in
relatively clear waters, whereas the TSM
and *Chl* determinations are numerous because they are made on the ship way. On the strength of that, the following procedure was applied to derive
regional bio-optical algorithms for the White Sea.

**Chlorophyll concentration**

For each water sample point, data of ocean color scanner MODIS-Aqua were collected, where available. Direct determination of concentration of phytoplankton pigments (chlorophyll “a” and pheophytin “a”) was performed by fluorometric method with seawater samples [25]. 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 (*R*_{rs}(531)/*R*_{rs}(547))^{-2.42}
n=68, *r*^{2}=0.61,
(5)

where
*R*_{rs}(531)
and *R*_{rs}(547)
are values of the above-surface remote-sensing reflectance.

**Particle backscattering and TSM**

To derive an algorithm for
calculation of TSM from satellite data, all data of ship measurements
of TSM in surface layer over the period of 2002-2010 were collected.
Each point of TSM measurements was considered as a potentially
sub-satellite one and data of ocean color scanners were collected for
each of them. Data corresponding to good weather conditions were used
in further analysis (also quality check of satellite data was carried
out). First,
the values of the particle backscattering coefficient *b*_{bp}(550)
were computed from MODIS-Aqua data (after
reprocessing 2009 and 2010) with
the algorithm described in Section __2.2__.

As a result, 195 pairs of quasi-concurrent
measurements of *b*_{bp}
and TSM were selected. Statistical analysis of these data led to the
following regression equation:

TSM=22.8 *b*_{bp}^{0.53}
*, n*=195,
*r*^{2}=0.70.
(6)

where
TSM is measured in mg/l, *b*_{bp}
– in m^{-1}.

**Yellow substance absorption**

To
derive values of the yellow substance absorption coefficient *a*_{g}_{
}, the
semi-analytic bio-optical algorithm was used [12]. 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 *R*_{rs}(*λ*_{i})
at six MODIS spectral bands 531, 547, 555, 645, 667, and 678 nm. The
values of *R*_{rs}(*λ*_{i})
were converted into values of the subsurface radiance reflectance
*ρ*(*λ*_{i})
by means of a formula given in [15].

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 *b*_{bp}.
and the yellow substance absorption coefficient *a*_{g}.