Search

Near-surface chlorophyll a concentration is a fundamental component of marine ecological processes, and its changes reflect the phytoplankton growth (primary productivity as well as loss due to grazing and sinking) feeding into higher trophic levels. Time series of measurements from several satellite sensors since late 1997 can be used as a proxy of chlorophyll a concentrations after calibrating against direct sea water measurements from oceanographic surveys. Previous studies indicate a need for a regional correction model in specific ‘case 2’ areas, where the relationship between satellite measurements and in situ measurements is different from the relationship in the general ‘case 1’ areas, due to complex environmental characteristics in different areas. Subarctic and boreal North Atlantic, including the waters around Iceland, have been considered case 2 waters, but a regional correction model has not been developed until now. We collated all relevant measurements of near-surface chlorophyll a from sea water samples, available in the Marine Research Institute database, and matched by date and location with satellite chlorophyll records, i.e. the GSM CHL1 records offered by the GlobColour Project. A multiple linear regression model was fitted to the observed in situ chlorophyll measurements, based on the satellite chlorophyll values (CHL1) and physical covariates: day of the year, sun elevation, and ocean depth. The resulting parsimonious model converts the satellite measurements to estimates that are in much better agreement with in situ measurements ( R 2 increases from 0.2 to 0.5), and is therefore proposed for calibration of regional corrections to the GlobColour Project’s GSM chlorophyll parameter, CHL1. [ABSTRACT FROM AUTHOR]

While remotely sensed data greatly improves the spatial and temporal resolution of ocean surface data, many data gaps still exist, particularly in northern latitude regions. We addressed this issue by reconstructing 8-day chl -a , sea surface temperature (SST) and photosynthetically available radiation (PAR) between 1998 and 2013. Direct matchups between the interpolated and in-situ chlorophyll -a data showed stronger correlations when both data types were combined although very little variation was found for the root mean square error across space and time. Chl -a shows a strong periodicity in the south and a weak periodicity in seasonal cycle in the north. Where periodicity is strong we found that correlations with the local environmental conditions were also the strongest. Wavelet coherence patterns showed that the phase correlation between SST, chl- a and NAO was dominated by an annual periodicity in the trends although the chl- a –NAO relationship was only significant after 2005. Ten regions were defined based on the k-means clustering of chl- a . Mean spring anomalies and the phenological timing of chl- a were defined for each region. The main drivers of spring anomalies were mean SST and PAR values while the spring timing was strongly correlated with the timing of MLD and PAR reaching threshold levels. [ABSTRACT FROM AUTHOR]

Guðmundsson, K., Heath, M. R., and Clarke, E. D. 2009. Average seasonal changes in chlorophyll a in Icelandic waters. – ICES Journal of Marine Science, 66: 2133–2140.The standard algorithms used to derive sea surface chlorophyll a concentration from remotely sensed ocean colour data are based almost entirely on the measurements of surface water samples collected in open sea (case 1) waters which cover ∼60% of the worlds oceans, where strong correlations between reflectance and chlorophyll concentration have been found. However, satellite chlorophyll data for waters outside the defined case 1 areas, but derived using standard calibrations, are frequently used without reference to local in situ measurements and despite well-known factors likely to lead to inaccuracy. In Icelandic waters, multiannual averages of 8-d composites of SeaWiFS chlorophyll concentration accounted for just 20% of the variance in a multiannual dataset of in situ chlorophyll a measurements. Nevertheless, applying penalized regression spline methodology to model the spatial and temporal patterns of in situ measurements, using satellite chlorophyll as one of the predictor variables, improved the correlation considerably. Day number, representing seasonal variation, accounted for substantial deviation between SeaWiFS and in situ estimates of surface chlorophyll. The final model, using bottom depth and bearing to the sampling location as well as the two variables mentioned above, explained 49% of the variance in the fitting dataset. [ABSTRACT FROM PUBLISHER]