Advanced Capabilities of Star Algorithm Processing Framework (Sapf) Version 2.0

The STAR Algorithm Processing Framework (SAPF) software is developed by the NOAA STAR Algorithm Scientific Software Integration and System Transition Team (ASSISTT) to facilitate and support transition of satellite remote sensing algorithms from research to operations. The SAPF is a single software framework with common interfaces for running and managing the precedence of large sets of enterprise algorithms needing to meet critical latency requirements. The ASSISTT Process Lifecycle and Algorithm Integration Team (PLAIT) integrates Level 2 product algorithms into the SAPF for both testing and operational implementation. The SAPF is currently run within NOAA Data Exploitation (NDE) to support the operational JPSS S-NPP and NOAA-20 VIIRS cloud, cryosphere, aerosol, and land products. In addition, it supports all GOES-16 baseline and enterprise algorithms as well as production generation for foreign satellites. SAPF version 1.0 is being replaced by version 2.0 in early 2019. Version 2.0 contains significant updates to support greater resource efficiency, scalability, improved run time performance, simplified configuration for users, as well as additional features such as parallel processing. The fully-featured FORTRAN and C/C++ Application Programming Interface (API) allows for seamless and easy integration of algorithms and other general software components in a multi-language programming environment. The highly scalable and flexible key-value based data architecture is designed to facilitate in-process and multi-processor data flows in a service-oriented environment and can be integrated with various data cache and data base technologies. The basic data transfer model allows for easy data exchange with other information systems. The XML based configuration allows efficient and user-friendly construction of multi-attribute data keys and effectively parameterizes and controls data flows in the system. The algorithm flow configuration supports and largely automates construction of any execution graphs, allowing for multi-satellite, multi-temporal, and otherwise varying configurations. The framework informs algorithms (via the API functions) which values have been requested allowing for selective and resource efficient algorithm execution. The fully automated and flexible input and output system transfers data between NetCDF files and the SAPF 2.0 data cache. The scene configuration and API support padded multi-dimensional and multi-resolution data input and segmentation description. These and other capabilities of SAPF 2.0 and how they support and enable the ASSISTT R2O process will be presented and discussed.