Image depicts SST data retrieved from TMI (microwave) and AVHRR (infrared) instruments northeast of Australia in August 2000.
The purpose of this study is to combine the relative strengths of infrared and microwave satellite SST observations so to produce an enhanced sea surface temperature product.
Merged data were created by averaging TMI and MCSST SST data after both products had been adjusted for biases due to diurnal warming of the surface layer and TMI biases caused by water vapor and wind roughened seas. The daily data have missing areas particularly in the subtropics. The weekly averaged merged data has no missing areas due to the increased satellite coverage during a 7-day period and spatial interpolation of the remaining missing areas.
The blended data uses optimal interpolation to determine daily SST fields using the MCSST and TMI SST data. Optimal interpolation method is based on the Reynolds and Smith (1994) method. Bias corrections were applied to both MCSST and TMI data as was done for the merged data. Daily SST fields having no missing data regions over the water.
Individual daily grid files are compressed into monthly tar files using gzip to ease the transfer of these file electronically. Tar files have the naming convention "NN.Yxxxx.Mxx.Vxx.type.tar.gz" where NN indicates the data type (bl indicates blended sst and mg indicates merged sst data), Y gives the four digit year, M the two digit month, V the 2 digit version number, type = sst for the sst data and type = pop for the population grids. Daily grid files have similar naming convention except they use Dxxx to identify the day number for the year and exclude the tar.gz ending.
The format of the data for all grid files is 2-byte INTEGERS. The array size of each grid is 1440 x 320. The data ranges all longitudes while latitude coverage is from 40S to 40N. The first Longitude index is centered on 0.125 degrees East, and the first latitude index is centered on 39.875S. A scale factor of 100 must be used for the SST data to achieve Celcius units. Population fields need no scaling. Missing data values are -9990 for the sst data (before scaling) and -999 for the population data.