Err...
As Prof Brian Cox said last time I saw him speak:
Observe (ie: take data points)
Try to understand the system you see
Try to build a model, reflecting your understanding
Repeat
It is critically important to build models, it's also critically important to use real world data.
Also, satellites need calibration, if you've got an accurate ground station you've got a datum to compare the satellite readings from.
And... There is no research money at the moment, NERC is being disbanded, which is a crying shame. The MetOffice have had their budgets slashed and this is coming out of research. Scientists are losing their jobs left right and centre.