A measurement collected today may become more useful when a future method is developed. That future depends on keeping the data, metadata and processing history interpretable.
Preserve what the next question may need
Focused Single Look Complex products store complex samples, retaining amplitude and phase for analyses such as interferometry. Detected Ground Range Detected products have undergone detection, multilooking and ground-range projection; they are useful for many intensity workflows but do not preserve the same complex phase information. Turning an SLC into an ordinary JPEG would discard numeric precision, complex phase, geospatial structure and much of the metadata needed for scientific analysis.
Scientific archives therefore separate image content from its description and lineage. Useful records include mission and product identifiers, acquisition time, orbit and geometry, polarisation, processing baseline, calibration state, coordinate reference, licence, quality flags and transformations. Versioned workflows record software, parameters and derived outputs so that another analyst can reproduce or challenge the path.
Formats and delivery choices
Complex raster formats and mission-native packages suit processing workflows; cloud-optimised geospatial formats can support efficient access to suitable derived rasters. JPEG 2000 may be appropriate for some raster products, but compression must be evaluated for the intended measurement and operation. Lossless storage preserves encoded values; lossy compression changes them. Wavelet methods can be useful, but there is no universal compression ratio that is scientifically safe for every workflow.
Chunking, tiling and cloud delivery can make large archives easier to query, while checksums and manifests help detect corruption. The architecture should keep source measurements distinct from calibrated data, derived products, model outputs and presentation images. Provenance is not administrative overhead: it is what lets tomorrow's method revisit today's evidence.