DON26TZ01-NV010 TITLE: Next Generation Tropical Cyclone Analysis, Forecasting, and Dissemination Tactical Decision Aid Software

OUSW (R&E) CRITICAL TECHNOLOGY AREA(S): Applied Artificial Intelligence (AAI)

COMPONENT TECHNOLOGY PRIORITY AREA(S): Advanced Computing and Software;Sustainment;Trusted AI and Autonomy

PROJECTED CMMC LEVEL REQUIREMENT: Level 2 (Self)

OBJECTIVE: Demonstrate an improved automated tropical cyclone forecasting, analysis, and dissemination tactical decision aid capability that uses a modern containerized software backend/frontend and is able to easily integrate legacy and novel component algorithms, models, databases, and Application Programming Interfaces (APIs).

DESCRIPTION: Domestic operational tropical cyclone forecasting at the Joint Typhoon Warning Center (DOW), Fleet Weather Centers (DOW), and National Hurricane Center (NOAA) have relied on the Automated Tropical Cyclone Forecast System (ATCF®) software suite for end-to-end tropical cyclone analysis, forecasting, and product dissemination for over three decades. This one-stop-shop for all data, modeling, post-processing, and user interaction for tropical cyclone information has endured due to its robust assured infrastructure, reliability, speed for executing actions, and long continuity even as forecasters and information have evolved. However, as compute environments and programming languages have changed, it has become more difficult to maintain and upgrade legacy software to take advantage of new capabilities.

This STTR topic seeks the development of a prototype software suite that can learn lessons from the success of ATCF®, but is architected in a modern software ecosystem to mitigate current workflow disadvantages. Fundamentally, the goal is a modular and containerized software application that can variously interact with legacy, current, and future software suites such as components of ATCF®, the Naval Integrated Tactical Environmental System Next Generation (NITES-Next) program, the NOAA Advanced Weather Interactive Processing System (AWIPS), and other back-end and front-end APIs. The software architecture must be designed from the outset to comply with DOW DevSecOps principles and prepare the system for the Risk Management Framework (RMF) process. Desired software requirements include design in a modern broadly supported and maintained open programming language that can run online or offline on premises or in a cloud compute environment; hardening against connectivity and bandwidth issues; separation of functionality between logic, database, analysis algorithms, forecast generation , user interface, and dissemination layers; modular component development where different parts can interoperate with other software; and the ability to quickly address software updates and functionality and revert on the client side.

A dual-pronged approach to improving workflow for the tropical cyclone forecast process is envisioned, with parallel development tracks for software architecture creation and decision support aid integration. While the focus of tropical cyclone tools in the 1980s through 2000s was on track and intensity prediction, the forecasting mission has increasingly expanded. This includes the ability to track before storms have formed; 3D storm structure and rainfall evolution; ocean and wave field information; storm surge and hazards; probabilistic uncertainty; and emerging machine learning tools. The software should be capable of generating and disseminating an automated, objectively optimal analysis and forecast product from available data — that is, without significant manual human effort. It is not expected that all these efforts are achievable within the scope of this STTR effort; however, the priority development schedule must be justified, and the software solution must be able to accommodate all of these components within a future strategic plan.

Back-end capability should include: (1) a modern development software framework that can easily include or remove proprietary and open source algorithms as desired and is built and deployed as a containerized architecture; (2) a robust state management (e.g., database) for storm information and aids with backwards compatibility and export for current ATCF® "deck file" formatting; (3) concurrency for data fetching and processing to reduce data latency for time-critical forecast workflows; and (4) defined APIs to provide data access to clients, such as front-end graphical user interfaces (GUI) or downstream machine-to-machine programs. Further, the back-end capability should support backup capability, likely through a distributed system of servers.

Front-end capability should include: (1) thin client(s) for use on desktop workstations and possibly within web browsers, facilitating both on-site and remote operation with both client-side and server-side rendering tested for DOW network responsiveness; (2) flexible means of calling external scripts/functions/APIs with configurable input data, allowing forecasters to trigger different production pipelines and workflows that could be defined externally; (3) editable runtime configuration facilitating separate profiles for different operational systems or users; (4) means to integrate with internal and/or external AI systems or agents to facilitate future workflows; and (5) GUI and storm management capabilities consistent with ATCF®, with additional emphasis on performant map navigation and rendering, multi-product and format overlays (from gridded and sparse data, such as from kml, kmz, ShapeFiles, GeoTiff, HDF, netCDF, GRIB2, Zarr, GeoJSON, etc.), and dynamic filtering and alerts for data as they populate in real time. It is imperative that software developed be done so with an emphasis that support does not require skills beyond those currently required and normally used by support staff at forecast centers.

PHASE I: Design and develop an architecture concept for an improved tropical cyclone software capability, identifying the most challenging technical components. Present a demonstration for viable solutions to the technological problems. Capabilities should be contextualized and contrasted with current ATCF functionality and forecasting requirements. Integration of emerging technologies should be explicitly described, and include machine learning prediction methods, higher resolution or local forecasting, and new machine learning tools to develop and publish a product. Required Phase I deliverables include the reporting documents on progress and outcomes, the final proposed architectural solution with justifications and risk/mitigation analysis, and a description/demonstration of the unique methods to address the development challenges.

PHASE II: Develop, demonstrate, and validate a prototype software suite. Effort should focus on proving back-end and front-end capabilities from the above topic description that address greatest needs from the concept developed in the Phase I. It is expected that regular engagement with Naval users and frequent software demonstrations and side-by-side comparison with other operational software suites will be performed throughout software maturation. The outcome by the end of the Phase II should be an end-to-end prototype (not expected to be feature complete) that addresses substantial aspects of the ATCF mission requirements and is able to be run robustly for real-time operational usage.

PHASE III DUAL USE APPLICATIONS: Focus on continuing development of software capability, while ramping up integration with other decision support tools in the operational environment. Performance equivalence with, or superiority to, ATCF is expected and should be demonstrated on different compute platforms, cloud systems, and classified systems. Performance metrics include: graphical rendering latency (both locally and over remote display protocols) and the time required for a user to complete a standard set of storm analysis and forecast tasks within the software. Acceptance of new data, analysis, and model capabilities from internal and external partners are expected. Awardee should be prepared to deploy as a stand-alone system with thin client interface, as a component backend in a thick client front end, and/or as a module in a multi-software system.

Beyond DOW, interaction and potential transition with NOAA is expected. There is potential for supporting multiple commercializations, such as supporting academic and basic open-source software community, software sales to commercial forecasting and risk companies, and further government capability development.

REFERENCES:

1. Qian, C, et al. "Tropical cyclone monitoring and analysis techniques: a review." Journal of Meteorological Research, 38(2), 13 May 2024, pp. 351-367. https://link.springer.com/article/10.1007/s13351-024-3135-9
2. Holbach, H. M. et al. "Recent advancements in aircraft and in situ observations of tropical cyclones." Tropical Cyclone Research and Review, 12(2), June 2023, pp. 81-99. https://www.sciencedirect.com/science/article/pii/S222560322300022X
3. Ricciardulli, L. et al "Remote sensing and analysis of tropical cyclones: Current and emerging satellite sensors." Tropical Cyclone Research and Review, 12(4), December 2023, pp. 267-293. https://www.sciencedirect.com/science/article/pii/S2225603223000553
4. Sampson, C. R. et al. "About ATCF." 12 April 2015. https://science.nrlmry.navy.mil/atcf/docs/ATCF-FAQ.html
5. Rappaport, E. N. et al "Advances and Challenges at the National Hurricane Center." Weather and Forecasting, 24(2), 1 April 2009, pp. 395-419. https://journals.ametsoc.org/view/journals/wefo/24/2/2008waf2222128_1.xml
6. Sampson, C. R. and Schrader, A. J. "The Automated Tropical Cyclone Forecasting System (Version 3.2)." Bulletin of the American Meteorological Society, 81(6), 2000, pp. 1231-1240. https://journals.ametsoc.org/view/journals/bams/81/6/1520-0477_2000_081_1231_tatcfs_2_3_co_2.pdf
7. Miller, R. J.; Schrader, A. J.; Sampson, C. R. and Tsui, T. L. "The automated tropical cyclone forecasting system (ATCF)." Weather and forecasting, 5(4), 01 December 1990, pp. 653-660. https://journals.ametsoc.org/view/journals/wefo/5/4/1520-0434_1990_005_0653_tatcfs_2_0_co_2.xml

KEYWORDS: Tropical cyclone; typhoon; Automated Tropical Cyclone Forecasting; ATCF; user interface; algorithms; modular; refactor; cloud; database; container; Application Programming Interface; API; graphical user interface; GUI; software

TPOC 1
Joshua Cossuth
joshua.h.cossuth.civ@us.navy.mil

TPOC 2
Levi Cowan
levi.p.cowan.civ@us.navy.mil