Storm Parameters Research Articles

Using surrogate modeling to predict storm surge on evolving landscapes under climate change

Planners managing coastal flood risk under a constrained computational budget face a tradeoff. Simulating many time periods or scenarios limits how many storm simulations can be run on each landscape. In this analysis, we present a deep learning model to predict storm surge as a function of storm parameters but also landscape features and boundary conditions (e.g., sea level). It is trained on peak surge elevations from Advanced Circulation (ADCIRC) hydrodynamic simulations of coastal Louisiana in a 2020 baseline and decadal periods from 2030 to 2070 under two morphological and climate scenarios. Leave-one-landscape-out cross-validation yielded a 0.086-m RMSE and 0.050-m MAE over 90 storms per landscape and 94,013 geospatial locations. A two-sided Kolmogorov-Smirnov test comparing annual exceedance probability (AEP) estimates from the model predictions to ADCIRC simulations rejected the null hypothesis that the predicted and ADCIRC AEP values were drawn from the same distribution only 1.1% of the time.

Statistics of significant storm events using one- and two-dimensional analyses of the natural and protected coasts of the Dziwnów Spit

In this paper, one- and two-dimensional statistical analyses of significant storm parameters were conducted along natural and protected coasts of the Dziwnów Spit, based on quantiles and concentration ellipses. In the one-dimensional analysis, the quantiles of 0.9, 0.99 and 0.999 erosion magnitude D, sea level F, significant wave height H and storm duration T were determined, and these quantiles correspond to significant storm occurrence once every 10, 100 and 1000 years, respectively. To account for the influence of other variables on the erosion magnitude, log-linear models describing the linear dependence of log(D) on F and log(D) on F and H were built. Based on these models, the corresponding quantiles for the erosion magnitude D were also determined. In the multivariate case, using the 2-dimensional normal distribution, (log(D), F), (log(D), H), and (log(D), T) concentration ellipses were determined for the above pairs of parameters for probabilities of 0.9, 0.99 and 0.999, respectively. The application of one-dimensional distribution results in the lowest values of eroded material of dune, while the use of concentration ellipses estimates the highest values of dune erosion. Moreover, the developed log-linear models better predict the values of eroded material of dune along natural coast than on protected one.

Storm's influence on long-term shoreline evolution along Casablanca-Mohammedia (Morocco)

Coastal erosion assessment has emerged as one of the most significant instruments for making decisions and offering effective solutions for coastal regions. Mohammedia-Casablanca's shoreline constantly changes due to natural and man-made erosion and accretion. This research uses historical aerial photographs, wave analysis data, geographic information system (GIS) tools, and statistical methods to evaluate spatial and temporal shoreline changes from 1969 to 2020. The analysis of storminess and rates of shoreline change is also discussed jointly in three sediment cells. For this, energy parameters are determined to characterize storm-induced impacts. The Digital Shoreline Analysis System (DSAS) calculated shoreline change using End Point Rate (EPR), Shoreline Change Envelope (SCE), and Net Shoreline Movement (NSM) methods. Out of 2477 transects placed at 30-meter intervals along the study area, approximately 39% of the transects show erosion, and the rest indicate accretion (61%). Over the entire study period (51 years), the analysis of shoreline evolution does not show a clear correlation with storm parameters. The only period in which the shoreline shift correlates with storm activity is 1969–1982. Storms were frequent and intense during this time, and the erosion was at its maximum (over 66% of the coastline). Outside this period, accretion has occurred even during periods of intense storms, as was the case here between 2009 and 2020, when accretion affected 70.1% of the Mohammadia-Casablanca coastline. This suggests that the recent installation of marine structures (groynes, seawalls, dikes, jetties, and piers) multiplied after 1982 has positively affected the rate of coastal erosion.

Distribution and recovery phase of geomagnetic storms during solar cycles 23 and 24

Abstract Coronal mass ejections (CMEs) and Stream Interaction Regions (SIRs) are the main drivers of intense geomagnetic storms. We study the distribution of geomagnetic storms associated with different drivers during solar cycles 23 and 24 (1996-2019). Although the annual occurrence rate of geomagnetic storms in both cycles tracks the sunspot cycle, the second peak in storm activity lags the second sunspot peak. SIRs contribute significantly to the second peak in storm numbers in both cycles, particularly for moderate to stronger-than-moderate storms. We note semiannual peaks in storm numbers much closer to equinoxes for moderate storms, and slightly shifted from equinoxes for intense and stronger-than-intense storms. We note a significant fraction of multiple-peak storms in both cycles due to isolated ICMEs/SIRs, while single-peak storms from multiple interacting drivers, suggesting a complex relationship between storm steps and their drivers. Our study focuses on investigating the recovery phases of geomagnetic storms and examining their dependencies on various storm parameters. Multiple-peak storms in both cycles have recovery phase duration strongly influenced by slow and fast decay phases with no correlation with the main phase buildup rate and Dst peak. However, the recovery phase in single-peak storms for both cycles depends to some extent on the main phase buildup rate and Dst peak, in addition to slow and fast decay phases. Future research should explore recovery phases of single and multiple-peak storms incorporating in-situ solar wind observations for a deeper understanding of storm evolution and decay processes.

Machine learning motivated data imputation of storm data used in coastal hazard assessments

In the Coastal Hazards System's (CHS) Probabilistic Coastal Hazard Analysis (PCHA) framework developed by the United States Army Corps of Engineers (USACE), historical records of tropical cyclone parameters have been used as data sources for statistical analysis, including fitting marginal distributions and measuring correlations between storm parameters. One limitation of the available historical databases is that observations of central pressure and radius of maximum winds are not available for a large number of storms. This may adversely affect the results of statistical analyses used to develop hazard curves. This study uses machine learning techniques to develop a data imputation method to “fill in” missing storm parameter records in historical datasets used for Joint Probability Method (JPM)-based coastal hazard analysis such as the USACE's CHS-PCHA. Specifically, Gaussian process regression (GPR) and artificial neural network (ANN) models are investigated as candidate machine learning-derived data imputation models, and the performance of different model parameterizations is assessed. Candidate imputation models are compared against existing statistical relationships. The effect of the data imputation process on statistical analyses (marginal distributions and correlation measures) is also evaluated for a series of example coastal reference locations.

A novel hybrid machine learning model for rapid assessment of wave and storm surge responses over an extended coastal region

Storm surge and waves are responsible for a substantial portion of tropical and extratropical cyclones-related damages. While high-fidelity numerical models have significantly advanced the simulation accuracy of storm surge and waves, they are not practical to be employed for probabilistic analysis, risk assessment or rapid prediction due to their high computational demands. In this study, a novel hybrid model combining dimensionality reduction and data-driven techniques is developed for rapid assessment of waves and storm surge responses over an extended coastal region. Specifically, the hybrid model simultaneously identifies a low-dimensional representation of the high-dimensional spatial system based on a deep autoencoder (DAE) while mapping the storm parameters to the obtained low-dimensional latent space using a deep neural network (DNN). To train the hybrid model, a combined weighted loss function is designed to encourage a balance between DAE and DNN training and achieve the best accuracy. The performance of the hybrid model is evaluated through a case study using the synthetic data from the North Atlantic Comprehensive Coastal Study (NACCS) covering critical regions within New York and New Jersey. In addition, the proposed approach is compared with two decoupled models where the regression model is based on DNN and the reduction techniques are either principal component analysis (PCA) or DAE which are trained separately from the DNN model. High accuracy and computational efficiency are observed for the hybrid model which could be readily implemented as part of early warning systems or probabilistic risk assessment of waves and storm surge.

STUDY ON THE TROPICAL-STORM-GENERATED WAVE FIELD ON THE SEA AREA AROUND LY SON ISLAND

Ly Son is a frontline island of our country that is often affected by adverse weather conditions such as monsoons or tropical storms. Studying the hydraulic parameters of tropical storms, including the storm's wave field, to determine extreme conditions plays an important role in forecasting, planning and construction of hydraulic works. The author uses the two-dimensional wave propagation model MIKE 21 SW to simulate the wave field of Molave storm 2020 in Ly Son sea. The results show a good agreement between the values obtained from the model and the actual measurements at Ly Son station. Based on the results obtained, the author proposes a method to simulate storm wave fields that is relatively accurate and effective. In particular, current standards based on wind speed and wind fetch still have many limitations. This is valuable in the actual design and planning of hydraulic works.

Efficacy of the Cytokine Adsorption Therapy in Patients with Severe COVID-19-Associated Pneumonia: Lesson Learned from a Prospective Observational Study

Introduction: Severe COVID-19 pneumonia can activate a cytokine storm. Hemoperfusion can reduce pro-inflammatory cytokines in sepsis but is still debated in the COVID-19 setting. Thus, we sought to investigate the benefits of HA-330 cytokine adsorption through clinical and laboratory outcomes. Methods: We conducted a single-center prospective observational study in adults with severe COVID-19 pneumonia admitted to the intensive care unit at Chiang Mai University Hospital (Chiang Mai, Thailand). Those with cytokine storms indicated by organ injury, including acute respiratory distress syndrome (ARDS), and high inflammatory markers were included. Patients treated with the HA-330 device were classified as a hemoperfusion group, while those without cytokine adsorption were classified as a control group. We compared the outcomes on day 7 after treatment and evaluated the factors associated with 60-day mortality. Results: A total of 112 patients were enrolled. Thirty-eight patients received hemoperfusion, while 74 patients did not. Baseline cytokine storm parameters were comparable. In univariate analysis, there was an improvement in clinical and laboratory effects from hemoperfusion therapy. In multivariate analysis, APACHE II score, SOFA score, PaO₂/FiO₂, the number of hemoperfusion sessions, the amount of blood purified, high-sensitivity C-reactive protein, and IL-6 were associated with mortality. Using at least 3 sessions of hemoperfusion could mitigate, the 60-day mortality (adjusted odds ratio 0.25, 95% confidence interval: 0.03–0.33, p = 0.001). By categorizing the amount of blood treated into 3 groups of <1 L/kg, 1–2 L/kg, and ≥2 L/kg, there was a linear dose-response association with survival, which was better in the higher volume purified (mortality 60% vs. 33.3% vs. 0%, respectively, p = 0.015). Conclusions: The early initiation of HA-330 hemoperfusion could improve the severity score and laboratory outcomes of COVID-19 ARDS. The optimal dose of at least three sessions or the amount of blood purified greater than 1 L/kg was associated with a reduction in 60-day mortality.

Impact of Storm Characteristics on Infiltration Dynamics in Sponge Cities Using SWMM

Effective stormwater management in urban areas requires enhancing the permeability of underlying surfaces. However, the impact of storm characteristics on infiltration processes in sponge cities remains insufficiently explored. This study uses the Horton method within the storm water management model to investigate how uniform and Chicago storm parameters affect infiltration rates. Our findings provide valuable insights: (1) Increasing porous pavement area proportionally reduces subarea sizes within subcatchments, and infiltration rates of porous pavements are supply-controlled. (2) Uniform storms result in consistent initial infiltration rates across pervious areas, subcatchments, and the entire catchment. The duration of this stable state decreases with higher return periods. Catchment infiltration volumes exhibit linear growth with greater storm intensities (R-squared = 0.999). (3) Peak infiltration rates and moments for pervious areas, subcatchments, and the overall catchment exhibit correlations with both the return period and the time-to-peak coefficient, with correlation coefficients ranging from −0.9914 to 0.9986 and p-values ranging from 0.0334 to 0.6923. This study quantifies the influence of design storm parameters on infiltration, providing valuable insights for stormwater infrastructure design and urban stormwater control.

Assessing North Atlantic Tropical Cyclone Rainfall Hazard Using Engineered-Synthetic Storms and a Physics-Based Tropical Cyclone Rainfall Model

Abstract In this study, we design a statistical method to couple observations with a physics-based tropical cyclone (TC) rainfall model (TCR) and engineered-synthetic storms for assessing TC rainfall hazard. We first propose a bias-correction method to minimize the errors induced by TCR via matching the probability distribution of TCR-simulated historical TC rainfall with gauge observations. Then we assign occurrence probabilities to engineered-synthetic storms to reflect local climatology, through a resampling method that matches the probability distribution of a newly proposed storm parameter named rainfall potential (POT) in the synthetic dataset with that in the observation. POT is constructed to include several important storm parameters for TC rainfall such as TC intensity, duration, and distance and environmental humidity near landfall, and it is shown to be correlated with TCR-simulated rainfall. The proposed method has a satisfactory performance in reproducing the rainfall hazard curve in various locations in the continental United States; it is an improvement over the traditional joint probability method (JPM) for TC rainfall hazard assessment.

Evaluation of the Storms Direct Runoff Prediction Methods used for Goizha-Dabashan Watershed

The Momentum and Aron &amp; White evaluating methods have been adopted to estimate the Nash Instantaneous Hydrograph parameters (IUH), while the two methods of excess rainfall (Ф-index and Natural Resources Conservation Service (NRCS) were applied in a model using a developed computer program in MATLAB to predict the direct runoff hydrograph for Goizha-Dabashan watershed located in the northeast of Iraq. In the verification stage, both Nash IUH optimal parameters of the storms and the average optimal values of the same parameters estimated in the calibration stage were applied and compared. The statistical tests showed a preference for the NRCS method with the momentum method in estimating direct runoff hydrograph (the average Nash-Sutcliffe Efficiency (NSE) was equal to 0.815 and 0.77 using optimal parameters verification storms and the average calibrated IUH parameters values, respectively). Also, satisfactory results (NSE was equal to 0.77 and 0.76 using storm parameters and the average calibrated IUH parameters values, respectively) were obtained by applying Aron &amp; White with the NRCS methods, which indicated the ability of both methods for estimating direct runoff hydrograph.

Evaluation of Convective Environments in the NARCliM Regional Climate Modeling System for Australia

Severe thunderstorms lead to multiple hazards including torrential precipitation, flash flood, hail, lightning, and wind gusts. The meso- to micro-scale nature of thunderstorms impose great challenges from understanding individual storm dynamics, storm climatology as well as projecting their future activities. High-resolution regional climate models can resolve the convective environments better than global models. Australia, especially the east and southeast parts of the continent, is a global hot spot for severe thunderstorms. This study evaluates the simulated convective environments from the New South Wales (NSW) and Australian Regional Climate Modelling (NARCliM) project based on the parameters of CAPE, CIN, 0–6-km vertical wind shear and storm severity. The ensemble regional downscaling is compared against the fifth-generation European Centre for Medium-range Weather Forecast Reanalysis (ERA5). The results show that although there are apparent biases (generally positive for CAPE and negative for CIN, and slightly overestimated vertical wind shear) in the downscaled storm parameters, the climatology of measures of storm severity over land, including their spatial patterns and seasonality, agree well with ERA5. These results have strong implication on the application of the climate projection to assess future convective environments in the region.

An Empirical Relationship among Characteristics of Severe Convective Storms, Their Cloud-Top Properties and Environmental Parameters in Northern Eurasia

Severe convective storms that produce tornadoes and straight-line winds usually develop under particular environmental conditions and have specific signatures on the cloud tops associated with intense updrafts. In this study, we performed a comparative analysis of satellite-derived characteristics, with a focus on cloud-top properties, and ERA5-based environmental parameters of convective storms in forested regions of the western part of Northern Eurasia in 2006–2021. The analyzed sample includes 128 different convective storms that produced 138 tornadoes and 143 linear windstorms. We found most tornadoes and linear windstorms are generated by quasi-linear convective storms or supercells. Such supercells form under lower convective instability and precipitable water content compared to those for other types of storms. We found a significant negative correlation of minimum temperature on the storm cloud top with instability parameters. In turn, the longevity of convective storms significantly correlates with wind shear and storm-relative helicity. About half of the tornadoes and 2/3 of linear windstorms are associated with the presence of cloud-top signatures, such as overshooting tops, cold-ring or cold U/V features. The events associated with such signatures are formed under high values of instability parameters. Our results can be used for further analysis of peculiarities of tornado and linear windstorm formation and to enhance the predictability of such severe events, especially in regions with a lack of weather radar coverage.

Large-scale circulation patterns and their influence on European winter windstorm predictions

Severe winter windstorms are amongst the most damaging weather events for Europe and show significant interannual variability. While surface variables (temperature, precipitation) have been successfully predicted for some time now, predictability of severe windstorms caused by extra-tropical cyclones remains less well explored. This study investigates windstorm prediction skill of the UK Met Office Global Seasonal Forecast System Version 5 (GloSea5) for the Northeast-Atlantic and European region. Based on an objective Lagrangian tracking of severe, damage relevant windstorms, three storm parameters are analysed: windstorm frequency and two intensity measures. Firstly, skill based on direct tracking of simulated windstorms is diagnosed. Significant positive skill for storm frequency and intensity is found over an extended area at the downstream end of the storm track, i.e., from the UK to southern Scandinavia. The skill for frequency agrees well with previous studies for older model versions, while the results of event-based intensity are novel. Receiver Operating Characteristic Curves for three smaller regions reveal significant skill for high and low storm activity seasons. Second, skill of windstorm characteristics based on their multi-linear regressions to three dominant large-scale circulation patterns [i.e., the North Atlantic Oscillation (NAO), the Scandinavian Pattern (SCA), and the East-Atlantic Pattern (EA)] are analysed. Although these large-scale patterns explain up to 80% of the interannual variance of windstorm frequency and up to 60% for intensity, the forecast skill for the respectively linear-regressed windstorms do not show systematically higher skill than the direct tracking approach. The signal-to-noise ratio of windstorm characteristics (frequency, intensity) is also quantified, confirming that the signal-to-noise paradox extends to windstorm predictions.

Estimation of beach erosion using Joint Probability analysis with a morphological model

Resilience of a beach is often defined by return period (RP) of a designed storm. During extreme events some correlation exists between storm parameters, such as wave height (Hs) and surge (S), and these parameters are also stochastic in nature. Thus, numerous combinations of Hs and S occurs for the same level of RP. The objective of this study is to demonstrate if a same level of joint RP with different combination of Hs and S would produce similar beach erosion, considering wave directions and orientation of coastline. To achieve this, a non-directional Joint Probability (JP) of storm Hs and S is integrated with a morphological model. Three scenarios along a JP contour are selected and erosions are estimated. The selected JP scenarios are converted into a synthetic storm time-series for morphological simulations. The results demonstrated that JP with a high Hs and low S scenario produced more erosion than other scenarios under constant direction. Regarding wave direction, for the same Hs and S, the shore-normal wave direction produced highest erosion. Thus, shore-normal wave with high Hs and low S produces the highest erosion than other combinations. This study demonstrates the importance of consideration various JP and wave directions in assessing resilience of a beach.

The effect of sequential storms on the performance of homogeneous berm breakwaters

Reshaping of homogeneous berm breakwaters have been experimentally investigated using 2D physical tests. Berm recession depends on storm duration in addition to breakwater geometry and other storm parameters such as significant wave height and peak period. Furthermore, a berm breakwater may experience some recessions prior to exposing to a design storm for which the breakwater is designed. In this study, 109 tests have been conducted totally to investigate the enhancement of berm recession during design storms and also to study the effect of sequential storms on the ultimate berm recession during the life span of berm breakwaters. The measured cumulative recessions have been compared with direct recessions obtained from a breakwater with no initial reshaping. According to the results, the effect of earlier sea states on final berm recession depends on the design stability index and it can be neglected only if the design storm is too weak ( H 0 T 0 < 40 ) or it is too strong ( H 0 T 0 > 80 ). For weak design storms, earlier reshaping is negligible, while for strong design storm, the final reshaped profile will be governed only by such design storms regardless of earlier sea states. The results show that the predicted berm recession against design storms can be maximum 10% less than the actual recession in the case of missing the effect of earlier sea states. Since this conclusion is valid for both statically and dynamically stable berm breakwaters (the maximum value occurs around H 0 T 0 = 70 ), it is suggested finally that designers consider 10% more recession than the predicted values by existing design formulae to be sure about the endurance of berm width against the cumulative berm recessions during design life of reshaping berm breakwaters. The observations showed also that this width can efficiently dissipate the wave energy and protect underneath layers. • Homogenous reshaping berm breakwaters are studied using 109 physical tests. • A new formula is proposed to calculate berm recession during storms as a function of modified stability index. • Cumulative recession during design life of berm breakwater is investigated and an applicable suggestion is provided. • A modified stability index is proposed for measuring the influence of design storms.

Does blood type have an effect on the course of COVID-19?

Introduction &#x0D; Predictive parameters that can affect the course of this infection have been the main topic of research since the beginning of the COVID-19 (Coronavirus disease 2019) pandemic. Since the discovery of blood groups, the effect of these on infectious diseases has always been of interest.&#x0D; Objectives &#x0D; To analyze the effect of ABO blood group on mortality, hospitalization duration and hematological and cytokine storm parameters in patients with COVID-19. &#x0D; Patients and methods: This retrospective study was conducted on 140 patients diagnosed with COVID-19. Demographic characteristics, laboratory parameters including ABO blood group, complete blood count (CBC) parameters, biochemical tests, cytokine storm parameters, duration of hospitalization, and final status (discharge or death) were recorded.&#x0D; Results: The 140 patients included in the analysis comprised 72 (51.4%) males and 68 (48.6%) females with a mean age of 66.3±14.0 years. . Age and gender, hospitalization duration and mortality rates were similar in all blood group types. Only D-dimer levels were found to be higher in blood group A compared with other blood groups.&#x0D; Conclusion: Although no difference in mortality was determined between groups, the D-dimer level was statistically significantly higher in COVID-19 patients with A blood group. Larger studies are needed to reflect D-dimer levels on the clinical course of infection, and thus on daily practice.

Convolutional Neural Network and Optical Flow for the Assessment of Wave and Tide Parameters from Video Analysis (LEUCOTEA): An Innovative Tool for Coastal Monitoring

Coastal monitoring is a topic continuously developing, which has been applied using different approaches to assess the meteo-marine features, for example, to contribute to the development of improved management strategies. Among these different approaches, coastal video monitoring coupled with recent machine learning and computer vision techniques has spread widely to assess the meteo-marine features. Video monitoring allows to obtain large spatially and temporally datasets well-distributed along the coasts. The video records can compile a series of continuous frames where tide phases, wave parameters, and storm features are clearly observable. In this work, we present LEUCOTEA, an innovative system composed of a combined approach between Geophysical surveys, Convolutional Neural Network (CNN), and Optical Flow techniques to assess tide and storm parameters by a video record. Tide phases and storm surge were obtained through CNN classification techniques, while Optical Flow techniques were used to assess the wave flow and wave height impacting the coasts. Neural network predictions were compared with tide gauge records. Furthermore, water levels and wave heights were validated through spatial reference points obtained from pre-event topographic surveys in the proximity of surveillance cameras. This approach improved the calibration between network results and field data. Results were evaluated through a Root Mean Square Error analysis and analyses of the correlation coefficient between results and field data. LEUCOTEA system has been developed in the Mediterranean Sea through the use of video records acquired by surveillance cameras located in the proximity of south-eastern Sicily (Italy) and subsequently applied on the Atlantic coasts of Portugal to test the use of action cameras with the CNN and show the difference in terms of wave settings when compared with the Mediterranean coasts. The application of CNN and Optical Flow techniques could represent an improvement in the application of monitoring techniques in coastal environments, permitting to automatically collect a continuous record of data that are usually not densely distributed or available.

Understanding Uncertainties in Tropical Cyclone Rainfall Hazard Modeling Using Synthetic Storms

Abstract Tropical cyclone (TC) rainfall hazard assessment is subject to the bias in TC climatology estimation from climate simulations or synthetic downscaling. In this study, we investigate the uncertainty in TC rainfall hazard assessment induced by this bias using both rain gauge and radar observations and synthetic-storm-model-coupled TC rainfall simulations. We identify the storm’s maximum intensity, impact duration, and minimal distance to the site to be the three most important storm parameters for TC rainfall hazard, and the relationship between the important storm parameters and TC rainfall can be well captured by a physics-based TC rainfall model. The uncertainty in the synthetic rainfall hazard induced by the bias in TC climatology can be largely explained by the bias in the important storm parameters simulated by the synthetic storm model. Correcting the distribution of the most biased parameter may significantly improve rainfall hazard estimation. Bias correction based on the joint distribution of the important parameters may render more accurate rainfall hazard estimations; however, the general technical difficulties in resampling from high-dimensional joint probability distributions prevent more accurate estimations in some cases. The results of the study also support future investigation of the impact of climate change on TC rainfall hazards through the lens of future changes in the identified important storm parameters.

Knowledge-Enhanced Deep Learning for Simulation of Extratropical Cyclone Wind Risk

Boundary-layer wind associated with extratropical cyclones (ETCs) is an essential element for posing serious threats to the urban centers of eastern North America. Using a similar methodology for tropical cyclone (TC) wind risk (i.e., hurricane tracking approach), the ETC wind risk can be accordingly simulated. However, accurate and efficient assessment of the wind field inside the ETC is currently not available. To this end, a knowledge-enhanced deep learning (KEDL) is developed in this study to estimate the ETC boundary-layer winds over eastern North America. Both physics-based equations and semi-empirical formulas are integrated as part of the system loss function to regularize the neural network. More specifically, the scale-analysis-based reduced-order Navier–Stokes equations that govern the ETC wind field and the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA) ERA-interim data-based two-dimensional (2D) parametric formula (with respect to radial and azimuthal coordinates) that prescribes an asymmetric ETC pressure field are respectively employed as rationalism-based and empiricism-based knowledge to enhance the deep neural network. The developed KEDL, using the standard storm parameters (i.e., spatial coordinates, central pressure difference, translational speed, approach angle, latitude of ETC center, and surface roughness) as the network inputs, can provide the three-dimensional (3D) boundary-layer wind field of an arbitrary ETC with high computational efficiency and accuracy. Finally, the KEDL-based wind model is coupled with a large ETC synthetic track database (SynthETC), where 6-hourly ETC center location and pressure deficit are included to effectively assess the wind risk along the US northeast coast in terms of annual exceedance probability.