Geometric Optimization

adapts the EnMAP image geometry to a given Sentinel-2 L2A dataset. Additionally the EnMAP co-registration and the keystone errors are taken into account. Fits the VNIR detector data to the reference image. Creates new L1B data with overlapping detetctors - problem no mosaicing due to 20 pixel loss).

Input

• L1B data
• DEM or mean altitude
• Sentinel-2 data sets
• global or local shifts
• cloud masks for EnMAP and S2

Output

• L1B data (coarse co-registered)
• Geometry layer (sensor geometry => UTM)
• Geometry layer (UTM => sensor geometry)

Memory Budget

Necessary Functions

• rcp2geolayer

Process Work Flow

• transform L1B to L1C using RPCs
• estimate coregistration error
• estimate keystone error
• estimate image shifts (VNIR)
• improves geometry layers