342
V. Barale and M. Gade
and MODIS-Aqua imagery, fully processed to correct top-of-the-atmosphere radiances from atmospheric noise; to derive normalized water-leaving radiances; and to
compute from these a score of derived parameters. In the web services’ version used
in the present case, the parameter chl is calculated using the OC4 empirical algorithm
for SeaWiFS data, and the similar OC3M algorithm for MODIS-Aqua data (O’Reilly
et al. 2000). Composite data products are obtained, for both SeaWiFS and MODIS,
by re-mapping on a common equal-area grid and then averaging with a grid cell,
or “bin”, ranging in size from 2 × 2 km
2 to 9 × 9 km
2 , depending on the web service used, at weekly or monthly intervals. Annual, multi-annual and climatological
composites can also be obtained, over the same grids. For each space/time bin, the
mean value (weighted by the number of valid pixels used in the binning process) is
accompanied by related data quality statistics (Campbell et al. 1995). The resulting
data products should be considered with caution, owing to the impact of various
coloured water constituents, or to direct bottom reflection, which may significantly
alter the chl signal derived from the sensor measurements, particularly in shallow
coastal areas. In fact, the presence of optically active materials other than phytoplankton and related pigments, with partially overlapping spectral signatures—i.e.
dissolved organic matter and suspended inorganic particles—can prevent the computation of reliable chl absolute values (Sathyendranath et al. 2000). However, when
these limitations are accounted for, the data can provide significant—if qualitative, at
worst—information on recurrent or anomalous algal blooms, and related boundary
conditions.
SeaWiFS data from ten consecutive full-year cycles, from July 1999 to June 2009,
were used to compute chl climatologies and chl average basin values (chl ABV ) from the
monthly composite images. Three different regions of interest were selected for the
spatial averaging: the entire basin (10
◦ N–30
◦ N, 32
◦ E–44
◦ E), the southern sub-basin
(10
◦ N–20
◦ N, 32
◦ E–44
◦ E) and the northern sub-basin (20
◦ N–30
◦ N, 32
◦ E–44
◦ E).
When the resulting SeaWiFS historical record proved to be incomplete (i.e. January,
February, March and July 2008; May 2009), the chl ABV time series was filled in using
MODIS-derived estimates. A statistical comparison of SeaWiFS and MODIS-Aqua
data, for the 7-year period of overlap between the two time series (i.e. July 2002
to June 2009), over the three regions considered, produced very high correlation
coefficients (about 0.9, for the entire basin case) of the chl ABV series generated by
each sensor. However, the linear regression of the two data sets indicated also that
the MODIS chl ABV are systematically larger of those computed from SeaWiFS data
by almost a factor of 2 (slope of 1.75, with a negative offset of 0.2, for the entire
basin case; see Fig. 17.2). Although this apparently systematic inconsistency was
not investigated further, it must be recalled that such uncertainties and biases are
not uncommon, when considering concurrent ocean colour products generated by
different orbital sensors (Zibordi et al. 2012).
Data collected by the microwave scatterometer SeaWinds, which was launched
on the QuikBird satellite in 1999 and operated until 2009, were used to derive wind
direction and speed statistics. The QuikBird SeaWinds, later dubbed QuikSCAT, was
the third in a series of National Aeronautic and Space Administration (NASA) scatterometers, operating at 13.4 GHz (Ku-band). Scatterometers transmit microwave
V. Barale and M. Gade
and MODIS-Aqua imagery, fully processed to correct top-of-the-atmosphere radiances from atmospheric noise; to derive normalized water-leaving radiances; and to
compute from these a score of derived parameters. In the web services’ version used
in the present case, the parameter chl is calculated using the OC4 empirical algorithm
for SeaWiFS data, and the similar OC3M algorithm for MODIS-Aqua data (O’Reilly
et al. 2000). Composite data products are obtained, for both SeaWiFS and MODIS,
by re-mapping on a common equal-area grid and then averaging with a grid cell,
or “bin”, ranging in size from 2 × 2 km
2 to 9 × 9 km
2 , depending on the web service used, at weekly or monthly intervals. Annual, multi-annual and climatological
composites can also be obtained, over the same grids. For each space/time bin, the
mean value (weighted by the number of valid pixels used in the binning process) is
accompanied by related data quality statistics (Campbell et al. 1995). The resulting
data products should be considered with caution, owing to the impact of various
coloured water constituents, or to direct bottom reflection, which may significantly
alter the chl signal derived from the sensor measurements, particularly in shallow
coastal areas. In fact, the presence of optically active materials other than phytoplankton and related pigments, with partially overlapping spectral signatures—i.e.
dissolved organic matter and suspended inorganic particles—can prevent the computation of reliable chl absolute values (Sathyendranath et al. 2000). However, when
these limitations are accounted for, the data can provide significant—if qualitative, at
worst—information on recurrent or anomalous algal blooms, and related boundary
conditions.
SeaWiFS data from ten consecutive full-year cycles, from July 1999 to June 2009,
were used to compute chl climatologies and chl average basin values (chl ABV ) from the
monthly composite images. Three different regions of interest were selected for the
spatial averaging: the entire basin (10
◦ N–30
◦ N, 32
◦ E–44
◦ E), the southern sub-basin
(10
◦ N–20
◦ N, 32
◦ E–44
◦ E) and the northern sub-basin (20
◦ N–30
◦ N, 32
◦ E–44
◦ E).
When the resulting SeaWiFS historical record proved to be incomplete (i.e. January,
February, March and July 2008; May 2009), the chl ABV time series was filled in using
MODIS-derived estimates. A statistical comparison of SeaWiFS and MODIS-Aqua
data, for the 7-year period of overlap between the two time series (i.e. July 2002
to June 2009), over the three regions considered, produced very high correlation
coefficients (about 0.9, for the entire basin case) of the chl ABV series generated by
each sensor. However, the linear regression of the two data sets indicated also that
the MODIS chl ABV are systematically larger of those computed from SeaWiFS data
by almost a factor of 2 (slope of 1.75, with a negative offset of 0.2, for the entire
basin case; see Fig. 17.2). Although this apparently systematic inconsistency was
not investigated further, it must be recalled that such uncertainties and biases are
not uncommon, when considering concurrent ocean colour products generated by
different orbital sensors (Zibordi et al. 2012).
Data collected by the microwave scatterometer SeaWinds, which was launched
on the QuikBird satellite in 1999 and operated until 2009, were used to derive wind
direction and speed statistics. The QuikBird SeaWinds, later dubbed QuikSCAT, was
the third in a series of National Aeronautic and Space Administration (NASA) scatterometers, operating at 13.4 GHz (Ku-band). Scatterometers transmit microwave