Sensor

Bit depth — 8-bit vs 16-bit

Determines how many distinct values the sensor can record per band. 16-bit preserves dynamic range needed for quantitative work.

8-BIT · 256 LEVELS16-BIT · 65,536 LEVELSshadowsaturationRADIOMETRIC RANGE · TYPICAL SCENE8-bit: clippedshadow + cloud8-bit is lossy for: SWIR spectral signatures · deep shadow detail · cloud edge analysis · change detection on subtle reflectance shifts
Fig. 1 Bit depth is how finely the sensor quantises measured radiance. 8-bit stores 256 levels; 16-bit stores 65,536. The difference is invisible in normal contrast but decisive in the dark and bright extremes — mineral exploration (SWIR), shadowed canyons, and saturated cloud edges all live in the range 8-bit throws away.

8-bit gives you 256 possible values per band. 16-bit gives you 65,536. 8-bit is fine for visualisation but throws away most of the sensor's dynamic range. 16-bit preserves the full range and is essential for quantitative work: reflectance, indices, time series, anything scientific.

When you need 16-bit

If you're computing NDVI, NDWI, or any other index where the absolute value matters, you need 16-bit. If you're doing time-series analysis, you need 16-bit. If you're a researcher doing radiative transfer modelling or biophysical inversions, you need 16-bit. We always deliver in the highest bit depth the sensor supports unless you specifically request 8-bit.