1) Satellites don't measure the same thing as ground thermometers (satellites measure the lower troposphere in its entirety, surface thermometers measure the surface temp) so it's not entirely unsurprising they don't give identical numbers to within an error margin.
2) Satellites don't measure polar temperatures very well, and polar amplification makes the temps at particularly the north pole go up faster than average - so a lower total response from the satellite data is expected.
3) People are freaking out over minor adjustments to the surface record which are well-supported by evidence (for example corrections made for a change in the time-of-day of the measurements at some stations at some point, or stations moving from city-centers to airports outside the city center) but the whole satellite record itself is full of far bigger corrections, the raw data of the satellites isn't a directly measured temperature but a remote sensor reading, which is influenced by a whole bunch of internal (sensor drift) and external (observation angle, satellite orbital height, weather conditions) factors. It's almost a miracle that they manage to get a useful data-series out of these satellite sensors.
Honestly, the only reason deniers try to argue that the satellite record is "more accurate" is because it shows a higher peak in 1998 and less total warming, allowing them to fiddle with the results to make it appear there's no warming at all - except that nowadays even the satellite record no longer supports that conclusion.