752 Processing Himawari Data through GOES-R Algorithms using the Algorithm Workbench Software Integration Suite

Wednesday, 13 January 2016
Erik Steinfelt, AER, Lexington, MA; and Y. He, D. Hogan, D. Hunt, R. J. Lynch, A. Werbos, and T. S. Zaccheo

The launch of Himawari-8 represents a significant milestone in meteorological remote sensing, offering greatly improved spatial, spectral and temporal observations over East Asia and the South Pacific. As Himawari data becomes more accessible, the availability of reusable multi-mission algorithms and software becomes increasingly more desirable. Many of the characteristics of these data are similar to those found in other systems, such as GOES-R, and the practice of redeveloping the entire code base for each new mission becomes inefficient and even impractical. In conjunction with our partners, we have developed and demonstrated a software interface solution that greatly mitigates the difficulties and expenses associated with deploying new ground processing software systems. In this work we provide an overview of this software implementation and demonstrate its applicability to provide customizable suites of Level 2 products (e.g. atmospheric, cloud, land and ocean products) that complement the impressive Himawari calibrated and geolocated Level 1b imagery. The Algorithm Workbench (AWB) is the AER implementation of this common interface. The superb resolution and quality of the data provides an ideal scenario under which to test the AWB tools. The AWB first acquires binary full disk AHI data, and decodes the data using common modular software elements, producing radiances and brightness temperature values in a simple HDF5 format. As part of the software suite's capabilities, the AHI data are imported along with other necessary data, and tested with operational GOES-R cloud algorithms. This process requires no code changes in the Algorithms or the infrastructure to operate on AHI data. The final products include cloud masks from ACM, cloud type and phase from ACT, and cloud top altitude, temperature, and pressure from ACH. By using the AWB in conjunction with the multi-mission algorithms, the effort to build a system that creates products from the Himawari data can be accomplished in a fraction of the time that was required for GOES-R development. Certain parameter calibrations were necessary to prepare the algorithms for execution. The AHI (Himawari) channels were mapped to ABI (GOES-R) channels, and provided as inputs to the AWB. Additionally, semi-static and ancillary inputs (NWP, Reynolds global SST, snow masks, etc.) were prepared for the associated observation times and sub-satellite position. The resulting imagery was familiar to the original GOES-R algorithm developers, making it easy to analyze and further calibrate the algorithm parameters to meet AHI instrument specifications. The science team performed an initial round of algorithm tuning to adjust algorithm parameters, subjectively optimizing the products. While a formal calibration/validation process is still required, the excellent initial results speak well of both the AHI instrument and the GOES-R cloud suite algorithms. By maintaining the same code base across multiple missions, algorithm improvements can be rapidly shared with users across the globe. The system is designed to be scalable, providing the ability to be run on a laptop or on a server, further enhancing the ability to share algorithms among varying user types. The current infrastructure is designed as a family of Python libraries, making it light weight and extremely flexible, while the algorithms themselves can be written in a variety of languages, including Python, C++ and Fortran. The entire system is of a modular design, allowing the user to write custom plug-in components where necessary. This presentation follows the process of adapting the Algorithm Workbench tool set and algorithms to create products from Himawari data. It provides validation for the excellent quality of the Himawari data and shows that the application of a common software approach is quite realistic. It describes and demonstrates the successful use of these tools and shows that great long-term advantage can be realized from early planning of genericized approaches and properly applied techniques.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner