PRISMA L2D Surface Reflectance

PRISMA L2D is the geocoded, orthorectified bottom-of-atmosphere surface reflectance product from the PRISMA hyperspectral mission operated by the Italian Space Agency (ASI)^[1][2]. The product covers 239 spectral bands across the VNIR-SWIR range from 400 to 2505 nm at 30 m ground sampling distance, distributed in HDF5 (.he5) format via the ASI CNM Access System at prisma.asi.it, free for registered users under ASI terms of use^[3].

L2D is the geocoded orthorectified version of L2C. Some literature treats L2C and L2D interchangeably^[1], while the formal product definition assigns L2C the georeferenced projection and L2D the fully orthorectified output^[2]. Users working with the archive should confirm which variant they have received.

The product organises reflectance into VNIR and SWIR HDF5 sub-cubes. Nine bands in the 920-1010 nm detector overlap zone appear in both sub-cubes; standard preprocessing retains VNIR-cube bands and discards the corresponding SWIR-cube bands in that zone before spectral analysis^[4]. Spectral regions near the detector boundary, approximately 885-970 nm and 1015-1050 nm, exhibit elevated noise and are routinely excluded in published preprocessing workflows^[5]. The atmospheric correction applied to produce L2D surface reflectance uses a MODTRAN-based algorithm; the processor version and correction parameters are not publicly documented by ASI as of the sources accessed^[1].

The instrument provides a 30 km swath width with a 29-day nadir repeat cycle. Off-nadir manoeuvring allows a revisit interval of approximately 7 days over targeted areas, though at reduced geometric quality relative to nadir acquisitions^[6]. PRISMA operates as a single satellite with no constellation redundancy, and acquisition gaps occur during instrument downtime or scheduling conflicts. The 30 m GSD limits sub-hectare mapping accuracy; outcrop-scale spectral mixing is common when characterising fractured or heterogeneous surfaces^[7].

Published applications include alteration mineral mapping and lithological discrimination across the full VNIR-SWIR range, carbonatite and rare-earth element deposit characterisation using SWIR absorption features, vegetation canopy biochemistry, land cover classification, and cross-sensor comparison against EnMAP L2A and EMIT L2A for geological mapping validation^[7][5].