Impact of irregular sampling by MERIS on eutrophication monitoring products for WFD and MSFD applications

The chlorophyll-a 90 percentile product (CHL-P90) gained importance in describing the eutrophication state of the Belgian part of the North Sea when Water Framework Directive requirements were transposed into Belgian law in July 2010. A MERIS based CHL-P90 product was developed to answer this need for information as MERIS provides a better temporal and spatial coverage compared to in-situ monitoring and thus is potentially better suited to provide a P90 estimate. However, the irregular availability of satellite chlorophyll-a observations both in space and time due to cloudiness, quality flagging, sensor malfunction, etc. has to be considered as it impacts the product accuracy. This impact is two-fold and dependent on (1) the availability of observations during the actual phytoplankton bloom and (2) a proportional distribution of observations in the bloom and non-bloom period. This effect of irregular sampling on the accuracy of CHL-P90 was studied in detail in simulations. The MIRO&CO-3D ecosystem model was used to generate realistic time series of chlorophyll-a concentrations with high temporal resolution in the most important Belgian monitoring stations. These time series were sub-sampled using the actual observation density of the MERIS satellite in Belgian waters. Results show that a mean relative error of 25.4% on the CHL-P90 estimate can be expected due to the effects of irregular sampling. The results of this study were used to improve the CHL-P90 algorithms by adding a weighing procedure taking into account irregular sampling issues to reduce these errors to 9.9%.