SatPy: A Python Library for Weather Satellite Processing

The amount of data sent to earth by meteorological satellites increases with each new satellite generation. These data are used by scientists, researchers, the interested public, and are assimilated to forecasting models, but become more difficult to use as technology allows for higher spatial and spectral resolutions. These datasets are usually provided to users in new file formats with multiple wavelengths, levels of calibration, polarization, and resolutions which can require new complicated software to read. Additionally, scientific analysis typically requires compositing various channels, reprojecting or interpolating, and writing to various file formats to suit the needs of the user. A new python library was created called SatPy for anyone who wants to work with satellite data and do it quickly and easily. SatPy is developed by a group of scientists and developers from around the world known as PyTroll; a group that has existed now for more than 7 years. SatPy is able to read over 30 different satellite data file formats from geostationary and polar-orbiting satellites and from various data sources. It provides multiple resampling algorithms, various RGB recipes that work across multiple sensors, and can write multiple image and data formats including GeoTIFF and NetCDF. SatPy provides common interfaces and data structures regardless of what data is being worked with and benefits from the dimension and coordinate handling of the XArray library. By leveraging the parallelization made available in the Dask library, SatPy is able to quickly process large arrays of satellite data that would normally not be possible on traditional user workstations. SatPy can be used as an interactive research tool or used as a stable operational processing library as many meteorological agencies and organizations are doing now. This talk will provide a detailed overview of SatPy’s features by first describing what use cases lead to its development and how SatPy accomplishes these historically complex tasks in only a few lines of Python. By providing all of this functionality and generating products in a free and open source software package we hope SatPy will become the go-to library for working with satellite data.