Accuracy of in situ sea surface temperatures used to calibrate infrared satellite measurements

The present computation of sea surface temperature (SST) from infrared sate llite measurements requires a coincident sample of in situ (drifting buoy a nd/or ship) SST measurements, to compute by regression the algorithmic coef ficients for the infrared data. Ignoring the fundamental difference between satellite-measured "skin SST" and buoy/ship measured "bulk SST," we analyz e past buoy and ship SST data to better evaluate the errors involved in the routine computation of SST from operational satellite data. We use buoy an d ship SST data for 2 years (1990 and 1996) from the Comprehensive Ocean-At mosphere Data Set as well as 2 years of previously cloud-cleared satellite radiances with matching drifting/moored buoy SST data from the NASA Pathfin der SST data set. We examine the in situ SST data for geographic distributi on, accuracy, and self-consistency. We find that there are large geographic regions that are frequently not sampled by the present drifting buoy netwo rk, a natural consequence of the fact that most buoys are not deployed to m easure in situ SST for satellite infrared SST calibration. Comparisons betw een drifting buoy SSTs suggest an error of similar to 0.4 degreesC for near ly coincident buoy SSTs. Comparing moored with adjacent drifting buoy SSTs, we find that drifting and moored buoy SSTs are samples from the same popul ation. Ship SSTs are noisier and have a significant warm bias relative to d rifting buoy SSTs. We explore the SST measurement accuracy changes that occ ur with variations in sampling coverage used for the SST algorithm regressi on. We both vary the total amount of points and restrict the regression dat a to regional sampling biases. Surprisingly the total number of calibration SST values can be quite small if they cover all latitudes. We conclude tha t buoy SSTs can have residual bias errors of similar to0.15 degreesC with R MS errors closer to 0.5 degreesC.