Processing levels (L1–L4)

Stages of data refinement. L1 = calibrated raw measurements. L2 = derived geophysical variables (e.g. temperature). L3 = gridded + time-averaged. L4 = model-assimilated or interpolated.

Why it matters

Most users want L2 or L3 — they’re scientist-ready. L1 needs heavy processing; L4 blends in a model so it’s not a pure observation. Always know which level you’re using.

Where you’ll meet it

• MODIS ships the full ladder: L1B calibrated radiances, L2 swath products like aerosol or SST, and tidy L3 daily/monthly global grids most users download.
• SMAP publishes L2 soil-moisture along the orbit and L3/L4 versions — the L4 product folds observations into a land model to give continuous, gap-free moisture and root-zone estimates.
• GPM IMERG is a classic L3 product: many sensors merged onto a common half-hourly global grid.
• GEDI lidar starts at L1B waveforms, then climbs to L2 canopy height and L4 aboveground biomass — a clear example of each level building on the one below.

In plain terms

Like coffee — L1 is raw beans, L2 is ground, L3 is brewed, L4 is a latte (the model is the milk that fills the gaps).