DEVOTES Science News – Estimation of the 90th percentile chlorophyll-a indicator using satellite data, in the south-east of the Bay of Biscay

By Y. Sagarminaga, M. Revilla, A. Borja, G. Rodríguez, G. Chust

The indicator “90th percentile of chlorophyll-a values over the defined growing season of a six year period” (hereafter named as P90-chla) is used under the Water Framework Directive (WFD), as one of the indicators within the phytoplankton element, to assess ecological status (high-good and good-moderate boundaries). The specific boundaries are defined through target chlorophyll concentrations/ranges set by the geographical intercalibration groups in Baltic Sea, North East Atlantic and Mediterranean. These target chlorophyll concentrations/ranges are determined locally for different water types and water categories, including coastal and transitional water bodies. The approach may be potentially applied for the Marine Strategy Framework Directive (MSFD) assessment, within Indicator “5.2.1 Chlorophyll concentration in the water column, under the D5 descriptor for Eutrophication”. Our objective was trying to demonstrate this.
This study estimated the P90-chla with in-situ and satellite data to compare the results of both methods and identify the sources of discrepancies and/or complementarities.
In-situ data were obtained from the Basque coast in the southeast Bay of Biscay, between 2003 and 2013 (Figure 1b).

Sagarminaga Fig.1

Figure 1. Location of the area of study. (a) Western Europe overlaid with the P90-chla calculated with the MODIS-Aqua OC5 dataset between 2003 and 2013. (b) Detailed information about the location of the sampled stations from the Basque Littoral Monitoring Network, overlaid with main rivers pouring into the coastal zone, WFD water bodies and the P90-chla calculated with the MODIS-Aqua OC5 dataset between 2003 and 2013.


Two space borne sensor datasets have been used: MODIS-Aqua data processed with OC5 algorithm and MODIS-Aqua processed with the OCI algorithm.
The results from this study show that satellite estimations of Chl-a are correlated with in-situ as shown in the yearly time series represented in Figure 2, although they overestimate the Chl-a values (OC5 algorithm overestimates less than the OCI algorithm). The R2 values from the linear fits of type II-SMA show higher values for stations with low river influence (0.41 for OC5 and 0.51 for OCI), but strongly decrease in stations with high river influence (0.05 for OC5 and 0.20 for OCI), indicating that non-algal particles and dissolved matter of terrestrial origin probably affect the estimation of Chl-a by the remote sensing bio-algorithms.

Sagarminaga Fig.2

Figure 2. Yearly chlorophyll-a geometric means estimated from in-situ (green line), MODIS-Aqua OC5 (blue line) and MODIS-Aqua OCI (orange line) datasets combined with the number of observations provided by each dataset (bars).

Consequently, the 6-year P90-chla provide higher values when calculated with satellite datasets and the classification results for the assessment are very different: in-situ dataset classifies the water in “high quality” class, whereas the OC5 dataset classifies them in “good quality” class and OCI in “poor quality” class (Figure 3).
These results indicate that in this area, the 6-year P90-chla indicator calculated using satellite data with the algorithms tested is only applicable in areas and/or periods with low suspended matter.

Sagarminaga Fig.3

Figure 3. Distributions of chlorophyll-a values estimated by the in-situ (left), MODIS-Aqua OC5 (middle) and MODIS-Aqua OCI (right) datasets using a 6-year running period as performed for the WFD assessment overlaid with the corresponding 90th percentile (red dot) and references values (colored horizontal lines).