Yearly averages of chlorophyll-a concentrations (in ng/cubic meter) are based on annual composites* of SeaWiFS satellite data provided by the SeaWiFS Project, NASA/Goddard Space Flight Center and GeoEye. Data were downloaded from the Level-3 Browser in the form of HDF files at a resolution of 9 x 9 km. Level-3 products are derived from true-color images generated from sub-sampled, calibrated, Rayleigh-corrected level-2 data, which are derived from raw radiance counts by applying sensor calibration, atmospheric corrections, and bio-optical algorithms. For more details on data format specifications, see the SeaWiFS product information page.
HDF files were converted to ESRI GRID format using the Marine Geospatial Ecology Toolbox, version 0.4. Coastline buffers were clipped by country and Exclusive Economic Zone (EEZ). A layer of EEZ boundaries was obtained from the VLIZ Maritime Boundaries Geodatabase (Version 1) of the Flanders Marine Institute, Belgium.
For defining coastal zones, we excluded the first ten kilometers of coastal waters because of the potential for bottom-reflectance or suspended sediments affecting the satellite measurements in close proximity to the coast. We limited the extent of coastal zones to 100 km offshore as an arbitrary cut-off above which impacts from land based sources on oceanic eutrophication are unlikely.
We converted radiance values into chlorophyll-a concentrations using the scaling equation provided in the HDF header files. We multiplied concentrations by 1,000 to obtain ng/cubic meter that could be stored in integer grids for further processing without losing valuable levels of accuracy. We exported attribute information from the clipped global grids representing near-coastal zones, and calculated the percentage of change between 1998 and 2007 for each grid cell (i.e., time series). To account for year-to-year fluctuations in chlorophyll concentrations, we assessed the strength of the relationship between time and chlorophyll concentration with a linear regression for each grid cell. Percentages of change were then expressed and mapped for those areas with significant changes in concentration only. Detailed documentation with step-by-step procedures and additional global statistics can be found in the following document:
Note: For regions with heavy seasonal cloud cover, there is a potential bias in the use of annual composites in that these composites make use of every cloud-free valid pixel in a year. Thus, less cloudy periods will be over-represented in the sample. If chlorophyll-a abundance is correlated with cloud cover in a given region - and it often is (e.g., coastal upwelling which brings phytoplankton to the surface in the Pacific Northwest also creates a cloud layer) - then the presence of clouds creates a sampling bias. This product does not address this issue, but in the future we plan to develop a product that does by using monthly composites instead of annual composites.
CIESIN/SEDAC would like to acknowledge the advice of Dr. Ajit Subramaniam of the Marine Biology Department of the Lamont-Doherty Earth Observatory at Columbia University in the development of these indicators. The data were developed by Steffen Foerster, Malanding Jaiteh, and Alex de Sherbinin.