Exploring the time and space differences between the sea surface temperature records from multiple satellite based IR and microwave sensors. Do we have a consistent record across scales of 1km to regional and global images?

For over three decades the most widely used atmospheric correction approach for deriving sea surface temperature (SST) in the 11-12μm has been the Non-Linear SST algorithm (NLSST). The NLSST form is used for both the NOAA Pathfinder SST, and the MODIS and VIIRS SST datasets produced by NASA. A variant is also the basis of the night-time algorithms using measurements in the mid-infrared window. Since 1981 ten IR sensors, with 4 different designs, and slightly different overpass times, are used to create a satellite based consistent record of SST. The AVHRR-2's were flown from 1981-1997, AVHRR-3's began in 1998, and in 2000 the MODIS era began. The most recent sensor, VIIRS, began collecting data in 2012 and begins a unique period with four polar orbiting IR sensors concurrently collecting SST, along with the measurements from the AMSR-2 microwave sensor. These dataset provide an opportunity to evaluate the consistency between SST from the different IR sensors in the 30-year record. A number of factors, relating to real geophysical differences in measurements, must be taken into account when making comparisons. These include depth of measurement, potential for diurnal heating, and differences in the time and space aggregation method used to create global files. Figure 1 shows the global daily 9km MODIS Aqua day and night SST relative to the Reynolds OI SST analysis (Panel A and B) and a composite microwave Windsat SST (Panel C). The comparison to the NCEP Reynolds OI, with a measurement depth referenced to buoys, suggests strong diurnal heating of the Arctic sea surface skin in the late summer. Presumably the diurnal signal is strongest in late summer due to more open ice free water, compared to spring and early summer when the day length is actually the longest of the year, but more ice may obscure the sea surface. Biases may also be present in the Reynolds field as fewer buoy observations at these latitudes guide the OI analysis. The day-time MODIS SST compared to Windsat SST (panel C), with a subsurface measurement depth of a few millimeters, is less warm relative to MODIS than the Reynolds fields, as might be expected,however the overall patterns are similar between the Windsat and Reynolds SST differences. Additional inter sensor and in situ comparisons and analysis for AVHRR, MODIS, VIIRS, and AMSR2 will be presented.