Hello, I am studying growth of filter feeding organisms (barnacles, mussels) in intertidal habitats along the Atlantic coast of Nova Scotia, Canada. I would like to use chlorophyll-a data from OceancolorWEB to get an estimate of phytoplankton abundance (food supply to these filter feeding organisms) at 5 locations along this coast. To do that, I downloaded MODIS-Aqua chl-a data and VIIRS chl-a data from OceancolorWEB. From both datasets, I chose mapped chl-a data (monthly, 4km, collected in April 2014) and extracted the chl-a data for the 5 locations using the pixel extraction tool in SeaDAS. I found that MODIS-Aqua and VIIRS chl-a data are quite similar in some locations (1,3,4), but not in others (2,5). Please see data (in mg chl-a /m3) below. Which of the two data sets can be recommended for intertidal habitats? Alternatively, would it be more precise to calculate the monthly mean chl-a concentrations by averaging the two monthly chl-a values from MODIS-Aqua and VIIRS?
Thank you very much
Location 1: 14.37758923 (MODIS), 12.26690292 (VIIRS)
Location 2: 1.530802011 (MODIS), 5.145525455 (VIIRS)
Location 3: 11.32187462 (MODIS), 14.34486485 (VIIRS)
Location 4: 10.33378124 (MODIS), 9.765091896 (VIIRS)
Location 5: 11.31599426 (MODIS), 6.451852322 (VIIRS)
Chl-a from MODIS-Aqua & VIIRS
Chl-a from MODIS-Aqua & VIIRS
Those sensors are far from ideal for intertidal locations. Two big issues for these sensors are:
1. land is highly reflective compared to water. Sensor responds drops off away from the pixel center, with a bell-shaped response. A bit of land near the edge of the "bell" can contribute a large portion of the total response even if the peak of the "bell" is more than a km. from the shore
2. in shallow water, there can also be contribution to the light reaching the sensor from the ocean floor. Bottom types vary widely, so the contribution to the light reaching the sensor will also vary.
The level-2 processing includes algorithms that attempt to mask out pixels contaminated by these effects, but these are somewhat sensor-specific due to differences in the
actual sensors. In general, ocean colour images tend to have unrealistically high chlor_ a values adjacent to shorelines. If there were very few days with clear sky in the month there will often be large differences in "monthly" averages between sensors dependiing on the timing of the satellite passes in relation to cloud cover.
If you are working in exposed areas with highly dynamic circulation, satellite pixels some distance from shore may better reflect chlor_a at the shore than pixels contaminated by the above effects. In other studies, time-series of 4km chlor_a and SST images for a region around the area of interest have proven useful in providing context (bloom timing, mixing events, etc.). Interpretation of such images can be a good exercise in physical and biological oceanography.
Level-2 data has better spatial resolution than the 4km monthly mapped files. Comparing regional semi-monthly composite images of chlor_a and SST for the two sensors you may well be able to explain the differences in monthly means. It may, however, be necessary to drill down to the indivdual level-2 files for a particular semi-monthly period where differences are large.
1. land is highly reflective compared to water. Sensor responds drops off away from the pixel center, with a bell-shaped response. A bit of land near the edge of the "bell" can contribute a large portion of the total response even if the peak of the "bell" is more than a km. from the shore
2. in shallow water, there can also be contribution to the light reaching the sensor from the ocean floor. Bottom types vary widely, so the contribution to the light reaching the sensor will also vary.
The level-2 processing includes algorithms that attempt to mask out pixels contaminated by these effects, but these are somewhat sensor-specific due to differences in the
actual sensors. In general, ocean colour images tend to have unrealistically high chlor_ a values adjacent to shorelines. If there were very few days with clear sky in the month there will often be large differences in "monthly" averages between sensors dependiing on the timing of the satellite passes in relation to cloud cover.
If you are working in exposed areas with highly dynamic circulation, satellite pixels some distance from shore may better reflect chlor_a at the shore than pixels contaminated by the above effects. In other studies, time-series of 4km chlor_a and SST images for a region around the area of interest have proven useful in providing context (bloom timing, mixing events, etc.). Interpretation of such images can be a good exercise in physical and biological oceanography.
Level-2 data has better spatial resolution than the 4km monthly mapped files. Comparing regional semi-monthly composite images of chlor_a and SST for the two sensors you may well be able to explain the differences in monthly means. It may, however, be necessary to drill down to the indivdual level-2 files for a particular semi-monthly period where differences are large.