An official website of the United States government
Abstract Because geostationary satellite (Geo) imagery provides a high temporal resolution window into tropical cyclone (TC) behavior, we investigate the viability of its application to short-term probabilistic forecasts of TC convective structure to subsequently predict TC intensity. Here, we present a prototype model that is trained solely on two inputs: Geo infrared imagery leading up to the synoptic time of interest and intensity estimates up to 6 h prior to that time. To estimate future TC structure, we compute cloud-top temperature radial profiles from infrared imagery and then simulate the evolution of an ensemble of those profiles over the subsequent 12 h by applying a deep autoregressive generative model (PixelSNAIL). To forecast TC intensities at hours 6 and 12, we input operational intensity estimates up to the current time (0 h) and simulated future radial profiles up to +12 h into a “nowcasting” convolutional neural network. We limit our inputs to demonstrate the viability of our approach and to enable quantification of value added by the observed and simulated future radial profiles beyond operational intensity estimates alone. Our prototype model achieves a marginally higher error than the National Hurricane Center’s official forecasts despite excluding environmental factors, such as vertical wind shear and sea surface temperature. We also demonstrate that it is possible to reasonably predict short-term evolution of TC convective structure via radial profiles from Geo infrared imagery, resulting in interpretable structural forecasts that may be valuable for TC operational guidance. Significance Statement This work presents a new method of short-term probabilistic forecasting for tropical cyclone (TC) convective structure and intensity using infrared geostationary satellite observations. Our prototype model’s performance indicates that there is some value in observed and simulated future cloud-top temperature radial profiles for short-term intensity forecasting. The nonlinear nature of machine learning tools can pose an interpretation challenge, but structural forecasts produced by our model can be directly evaluated and, thus, may offer helpful guidance to forecasters regarding short-term TC evolution. Since forecasters are time limited in producing each advisory package despite a growing wealth of satellite observations, a tool that captures recent TC convective evolution and potential future changes may support their assessment of TC behavior in crafting their forecasts.
more »
« less
This content will become publicly available on June 1, 2026
Tropical Cyclone Surface Winds From Aircraft With a Neural Network
Estimates of the surface wind field in a tropical cyclone (TC) are required in real time by operational forecast centers to warn the public about potential impacts to life and property. In‐situ aircraft data must be adjusted from flight level to surface using wind reductions (WRs) since the aircraft cannot fly too low due to safety concerns. Current operational WRs do not capture all the variability in the TC surface wind field. In this study, an observational data set of Stepped Frequency Microwave Radiometer (SFMR) surface wind speeds that are collocated with flight‐level predictors is used to analyze the variability of WRs with respect to aircraft altitude and TC storm motion and intensity. The Surface Winds from Aircraft with a Neural Network (SWANN) model is trained on the observations with a custom loss function that prioritizes accurate prediction of relatively rare high‐wind observations and minimization of variance in the WRs. The model is capable of learning physical relationships that are consistent with theoretical understanding of the TC boundary layer. Radar‐derived wind fields at flight level and independent dropwindsonde in‐situ surface wind measurements are used to validate the SWANN model and show improvement over the current operational procedure. A test case shows that SWANN can produce a realistic asymmetric surface wind field from a radar‐derived flight‐level wind field which has a maximum wind speed similar to the operational intensity, suggesting promise for the method to lead to improved real‐time TC intensity estimation and prediction in the future.
more »
« less
- Award ID(s):
- 2103776
- PAR ID:
- 10649859
- Publisher / Repository:
- Journal of Geophysical Research: Machine Learning and Computation,
- Date Published:
- Journal Name:
- Journal of Geophysical Research: Machine Learning and Computation
- Volume:
- 2
- Issue:
- 2
- ISSN:
- 2993-5210
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Properties of frozen hydrometeors in clouds remain difficult to sense remotely. Estimates of number concentration, distribution shape, ice particle density, and ice water content are essential for connecting cloud processes to surface precipitation. Progress has been made with dual-frequency radars, but validation has been difficult because of lack of particle imaging and sizing observations collocated with the radar measurements. Here, data are used from two airborne profiling (up and down) radars, the W-band Wyoming Cloud Radar and the Ka-band Profiling Radar, allowing for Ka–W-band dual-wavelength ratio (DWR) profiles. The aircraft (the University of Wyoming King Air) also carried a suite of in situ cloud and precipitation probes. This arrangement is optimal for relating the “flight-level” DWR (an average from radar gates below and above flight level) to ice particle size distributions measured by in situ optical array probes, as well as bulk properties such as minimum snow particle density and ice water content. This comparison reveals a strong relationship between DWR and the ice particle median-volume diameter. An optimal range of DWR values ensures the highest retrieval confidence, bounded by the radars’ relative calibration and DWR saturation, found here to be about 2.5–7.5 dB. The DWR-defined size distribution shape is used with a Mie scattering model and an experimental mass–diameter relationship to test retrievals of ice particle concentration and ice water content. Comparison with flight-level cloud-probe data indicate good performance, allowing microphysical interpretations for the rest of the vertical radar transects.more » « less
-
Accurate specification of hurricane inner-core structure is critical to predicting the evolution of a hurricane. However, observations over hurricane inner cores are generally lacking. Previous studies have emphasized Tail Doppler radar (TDR) data assimilation to improve hurricane inner-core representation. Recently, Doppler wind lidar (DWL) has been used as an observing system to sample hurricane inner-core and environmental conditions. The NOAA P3 Hurricane Hunter aircraft has DWL installed and can obtain wind data over a hurricane’s inner core when the aircraft passes through the hurricane. In this study, we examine the impact of assimilating DWL winds and TDR radial winds on the prediction of Hurricane Earl (2016) with the NCEP operational Hurricane Weather Research and Forecasting (HWRF) system. A series of data assimilation experiments are conducted with the Gridpoint Statistical Interpolation (GSI)-based ensemble-3DVAR hybrid system to identify the best way to assimilate TDR and DWL data into the HWRF forecast system. The results show a positive impact of DWL data on hurricane analysis and prediction. Compared with the assimilation of u and v components, assimilation of DWL wind speed provides better hurricane track and intensity forecasts. Proper choices of data thinning distances (e.g., 5 km horizontal thinning and 70 hPa vertical thinning for DWL) can help achieve better analysis in terms of hurricane vortex representation and forecasts. In the analysis and forecast cycles, the combined TDR and DWL assimilation (DWL wind speed and TDR radial wind, along with other conventional data, e.g., NCEP Automated Data Processing (ADP) data) offsets the downgrade analysis from the absence of DWL observations in an analysis cycle and outperforms assimilation of a single type of data (either TDR or DWL) and leads to improved forecasts of hurricane track, intensity, and structure. Overall, assimilation of DWL observations has been beneficial for analysis and forecasts in most cases. The outcomes from this study demonstrate the great potential of including DWL wind profiles in the operational HWRF system for hurricane forecast improvement.more » « less
-
Abstract The impacts of a tropical cyclone after landfall depend not only on storm intensity but also on the size and structure of the wind field. Hence, a simple predictive model for the wind field after landfall has significant potential value. This work tests existing theory for wind structure and size over the ocean against idealized axisymmetric landfall experiments in which the surface beneath a mature storm is instantaneously dried and roughened individually or simultaneously. Structure theory captures the response of the low-level wind field to different types of idealized landfalls, given the intensity and size response. Storm size, modeled to follow the ratio of simulated time-dependent storm intensity to the Coriolis parameter, can generally predict the transient response of the storm gale wind radiir34ktto inland surface forcings, particularly for at least moderate surface roughening regardless of the level of drying. Given knowledge of the intensity evolution, the above results combine to yield a theoretical model that can predict the full tangential wind field response to idealized landfalls. Significance StatementA theoretical model that can predict the time-dependent wind field structure of landfalling tropical cyclones (TCs) with a small number of physical, observable input parameters is essential for mitigating hazards and allocating public resources. This work provides a first-order prediction of storm size and structure after landfall, which can be combined with existing intensity predictions to form a simple model describing the inland wind field evolution. Results show its potential utility for modeling idealized inland TC wind fields.more » « less
-
Vertical eddy diffusivity (VED) in the planetary boundary layer (PBL) has a significant impact on forecasts of tropical cyclone (TC) structure and intensity. VED uncertainties in PBL parameterizations can be partly attributed to the model’s inability to represent roll vortices (RV). In this study, RV effects on turbulent fluxes derived from a large eddy simulation (LES) by Li et al. (Geophys. Res. Lett., 2021, 48, e2020GL090703) are added to the VED parameterization of the PBL scheme within the operational Hurricane Weather Research and Forecasting (HWRF) model. RV contribution to VED is parameterized through a coefficient and varies with the RV intensity and velocity scale. A modification over land has also been implemented. This modified VED parameterization is compared with the original wind-speed-dependent VED scheme in HWRF. Retrospective HWRF forecasts of Hurricanes Florence (2018) and Laura (2020) are analyzed to evaluate the impacts of the modified VED scheme on landfalling hurricane forecasts. Results show that the modified PBL scheme with the RV effect leads to an improvement in 10-m maximum wind speed forecasts of 14%–31%, with a neutral to positive improvement for track forecasts. Improved wind structure and precipitation forecasts against observations are also noted with the modified PBL scheme. Further diagnoses indicate that the revised PBL scheme enhances moist entropy in the boundary layer over land, leading to improved TC intensity prediction compared to the original scheme.more » « less
