Implementing Forecaster Support Products for Analysis of Tropical Cyclone Intensity and Structure from Aircraft Reconnaissance Observations

Observations in tropical cyclones (TCs) from aircraft-based radar, dropsondes, and reconnaissance flight level and surface wind measurements often seem to conflict due to the disparate spatial and temporal sampling characteristics between instrument types, the representativeness of an observation on the vortex-scale, and systematic or random observational errors. Forecasters working in a time-limited environment must integrate the available information using various adjustments, and even then, apparent conflicts often remain between observational sources. Substantial research has examined the relationships between measurements from various instruments and how these relate to TC maximum sustained winds (VMAX), radius of maximum winds (RMW), and size, yet the amount of data that can be brought to bear on this topic has grown considerably since these studies were conducted. Meanwhile, analysis techniques have continued to advance, offering significant potential to improve the forecast analysis process.

This 3-year project seeks to operationally implement aircraft reconnaissance products to optimize the spatial and temporal analysis of the surface wind field in TCs. The main project outcomes will be a real-time suite of observational analysis products that provide high quality spatial and temporal analysis of the TC wind field from 0-3 km in height, graphical and tabular outputs of the estimated VMAX, RMW, and wind radii information along with the uncertainty associated with the estimates. The products will allow forecasters to analyze TC wind structure more quickly and with greater confidence. This presentation will provide an overview of the capabilities which are being implemented, including:

• a near-real-time data stream of Vortex Data Messages (VDM-RT),
• a near-real-time data stream of the flight level observations (FLIGHT-RT), with on-the-fly center finding and radial leg parsing,
• a new machine-learning-based method for reducing flight level winds to the surface,
• near-real-time 2-d and 3-d analyses based on a lightweight version of Spline Analysis at Mesoscale Utilizing Radar and Aircraft Instrumentation (SAMURAI), and
• an hourly temporal analysis based on a one-dimensional variational (1-d Var) data assimilation approach that objectively blends data based on the representativeness and error characteristics of each observation.

The resulting real-time suite of observational analysis products are intended to provide high quality spatial and temporal analysis of the TC wind field. Specific products will include graphical and tabular outputs of the estimated VMAX, RMW, and wind radii information along with uncertainty bounds and the analyzed TC wind field from the surface up to ~3 km height. By demonstrating the analysis products in an operational environment in the upcoming 2024 and 2025 hurricane seasons, this project aims to reduce NHC forecaster workload and improve forecasters’ confidence. The analysis products should also lead to improvements to hurricane track, structure, and intensity forecasts.

Figure caption: Wind ratios from the FLIGHT+ neural network training dataset showing asymmetry in observed wind ratios, stratified by intensity: tropical storm (TS), hurricane (HU), and major hurricane (MH). The errors have different characteristics at different intensities.