SarPy

SarPy is a basic Python library to read, write, and do simple processing of complex SAR data using the NGA SICD format (standards linked below). It has been released by NGA to encourage the use of SAR data standards throughout the international SAR community. SarPy complements the SIX library (C++) and the MATLAB SAR Toolbox, which are implemented in other languages but have similar goals.

Some sample SICD files can be found here.

Relevant Standards Documents

A variety of SAR format standard are mentioned throughout this ReadMe, here are associated references.

Sensor Independent Complex Data (SICD) - latest version (1.2.1; 2018-12-13)

1. Volume 1, Design & Implementation Description Document
2. Volume 2, File Format Description Document
3. Volume 3, Image Projections Description Document
4. Schema

Sensor Independent Derived Data (SIDD) - latest version (2.0; 2019-05-31)

1. Volume 1, Design and Implementation Description Document
2. Volume 2, NITF File Format Description Document
3. Volume 3, GeoTIFF File Format Description Document
4. Schema

Compensated Phase History Data (CPHD) - latest version (1.0.1; 2018-05-21)

1. Design & Implementation Description
2. Design & Implementation Schema

Both SICD and SIDD files are NITF files following specific guidelines National Imagery Transmission Format (NITF) - latest version (2.1, Revision C; 2017-06-06)

1. National Imagery Transmission Format

For other NGA standards inquiries, the standards registry can be searched here.

Basic Capability

The basic capabilities provided in SarPy is generally SAR specific, and largely geared towards reading and manipulating data provided in NGA SAR file formats. Full support for reading and writing SICD, SIDD, CPHD, and CRSD (standard pending) and associated metadata structures is currently provided, and this is the main focus of this project.

There is additionally support for reading data from complex data formats analogous to SICD format, usually called Single Look Complex (SLC) or Level 1, from a variety of commercial or other sources including

• Capella (partial support)
• COSMO-SkyMed (1st and 2nd generation)
• GFF (Sandia format)
• ICEYE
• NISAR
• PALSAR2
• RadarSat-2
• Radar Constellation Mission (RCM)
• Sentinel-1
• TerraSAR-X.

For this SLC format data, it is read directly as though it were coming from a SICD file. This ability to read does not generally apply to data products other than the SLC or Level 1 product, and there is typically no direct NGA standard analog for these products.

Some general TIFF and NITF reading support is provided, but this is not the main goal of the SarPy library.

Documentation

Documentation for the project is available at readthedocs.

If this documentation is inaccessible, it can be built locally after checking out this repository using sphinx via the command python setup.py build_sphinx. This depends on python packages sphinx and sphinxcontrib-napoleon.

Origins

SarPy was developed at the National Geospatial-Intelligence Agency (NGA). The software use, modification, and distribution rights are stipulated within the MIT license.

Dependencies

The core library functionality depends only on numpy >= 1.11.0 and scipy.

Optional Dependencies and Behavior

There are a small collection of dependencies representing functionality which may not be core requirements for much of the sarpy targeted tasks. The tension between requiring the least extensive list of dependencies possible for core functionality and not having surprise unstated dependencies which caused unexpected failures is evident here. It is evident that there are many viable arguments for making any or all of these formally stated dependencies. The choices made here are guided by practical realities versus what is generally considered best practices.

For all packages on this list, the import is tried (where relevant), and any import errors fr these optional dependencies are caught and handled. In other words, a missing optional dependency will not be presented as import time. Excepting the functionality requiring h5py, this import error handling is probably silent.

Every module in sarpy can be successfully imported, provided that numpy and scipy are in the environment. Attempts at using functionality depending on a missing optional dependency will generate an error at run time with accompanying message indicating the missing optional dependency.

• Support for reading single look complex data from certain sources which provide data in hdf5 format require the h5py package, this includes Cosmo-Skymed, ICEYE, and NISAR data.

• Reading an image segment in a NITF file using jpeg or jpeg 2000 compression and/or writing a kmz image overlay requires the pillow package.

• CPHD consistency checks, presented in the sarpy.consistency module, depend on lxml>=4.1.1, networkx>=2.5, shapely>=1.6.4, and pytest>=3.3.2. Note that these are the versions tested for compliance.

• Some less commonly used (in the sarpy realm) NITF functionality requires the use and interpretation of UTM coordinates, and this requires the pyproj package.

• Building sphinx documentation (mentioned below) requires packages sphinx, sphinxcontrib-napoleon, and sphinx_gallery.

• Optional portions of running unit tests (unlikely to be of relevance to anyone not performing development on the core sarpy package itself) require the lxml package

Installation

From PyPI, install using pip (may require escalated privileges e.g. sudo):

pip install sarpy

Note that here pip represents the pip utility for the desired Python environment.

For verbose instructions for installing from source, see here. It is recommended that still the package is built locally and installed using pip, which allows a proper package update mechanism, while python setup.py install does not.

Issues and Bugs

Support for Python 2 has been dropped. The core sarpy functionality has been tested for Python 3.6, 3.7, 3.8, 3.9, and 3.10.

Changes to sarpy for the sole purpose of supporting a Python version beyond end-of-life are unlikely to be considered.

Information regarding any discovered bugs would be greatly appreciated, so please feel free to create a github issue. If more appropriate, do not hesitate to contact thomas.mccullough.ctr@nga.mil for assistance.

Pull Requests

Efforts at direct contribution to the project are certainly welcome, and please feel free to make a pull request. Note that any and all contributions to this project will be released under the MIT license.

Software source code previously released under an open source license and then modified by NGA staff is considered a "joint work" (see 17 USC 101); it is partially copyrighted, partially public domain, and as a whole is protected by the copyrights of the non-government authors and must be released according to the terms of the original open source license.

Associated GUI Capabilities

Some associated SAR specific graphical user interface tools are maintained in the sarpy_apps project.