What good is real data mixed with model results unless the real data is refusing to cooperate with a foregone conclusion?
One hundred years ago, ships typically lowered a bucket in the sea and then measured the temp once brought back onboard; how has the measurement technique changed since?
Most ships currently measure the temperature of the surface intake water. But now there is also an increasing number of drifters returning both subsurface and surface temperatures via satellite uplink.
There are a number of references available evaluating the techniques, constructing time-series based on various methodologies, corrections, etc.