Data assimilation

Blending observations into a physical model so the model state stays consistent with reality. SMAP L4 and MERRA-2 are assimilation products.

Why it matters

Assimilation fills the gaps satellites miss (in time and space) with physics, producing complete, continuous fields — at the cost of being part-model, not pure observation.

In plain terms

Letting real measurements continually “nudge” a simulation back on track — like GPS correcting a map app’s estimate of where you are.