Satellite imagery and cal/val reference data, through one Python interface.
spectrAccess is an open-source Python library that finds and downloads Earth observation satellite imagery, and the ground and atmospheric reference data used to calibrate and validate it, through one interface. It returns each provider's data as published, with nothing corrected or recalibrated on the way.
spectrAccess fetches from the space agencies, archives and measurement networks that publish the data. Every source works the same way: find what is available, download it, read it. Tables come back as pandas DataFrames and gridded products as xarray Datasets. Portal logins, file formats and unit conventions are handled in the package, so the time goes into the analysis.
It does not correct, calibrate or reprocess anything. The provider's metadata, quality flags and uncertainty are kept wherever the provider supplies them, so your processing starts from the source.
Where a source supports it, measurements can also be returned in one shared schema. There, uncertainty is a record rather than a bare number: the published uncertainty where the source gives one, and a plain statement where it does not. Each value keeps its source, URL and retrieval time, so a result can be traced back to what was downloaded and when.
Anyone who works with satellite imagery or the reference data behind it and wants both from one place: Earth observation and remote sensing scientists, calibration and validation engineers, data engineers, satellite data providers and the analysts who build on their data.
pip install spectraccess Optional extras and a short example are on the PyPI page.
spectrAccess fetches data on your behalf, using your own account where a provider needs one. It never re-serves or redistributes source data, and each connector lists its provider's terms and citation requirements.
Data clients stop working quietly: portals move, formats change, and an unmaintained connector returns nothing without saying why. RefCal, our cross-sensor calibration layer, reads its imagery and reference data through these same connectors, so keeping them working is part of our own work. We publish them so others do not have to rebuild the same plumbing.
Using spectrAccess?
Tell us which source you need next, or where a connector falls short. Issues are also welcome on GitHub.