An official website of the United States government
Abstract Tropical cyclones (TCs) are one of the greatest threats to coastal communities along the US Atlantic and Gulf coasts due to their extreme wind, rainfall and storm surge. Analyzing historical TC climatology and modeling TC hazards can provide valuable insight to planners and decision makers. However, detailed TC size information is typically only available from 1988 onward, preventing accurate wind, rainfall, and storm surge modeling for TCs occurring earlier in the historical record. To overcome temporally limited TC size data, we develop a database of size estimates that are based on reanalysis data and a physics‐based model. Specifically, we utilize ERA5 reanalysis data to estimate the TC outer size, and a physics‐based TC wind model to estimate the radius of maximum wind. We evaluate our TC size estimates using two high‐resolution wind data sets as well as Best Track information for a wide variety of TCs. Using the estimated size information plus the TC track and intensity, we reconstruct historical storm tides from 1950 to 2020 using a basin‐scale hydrodynamic model and show that our reconstructions agree well with observed peak storm tide and storm surge. Finally, we demonstrate that incorporating an expanded set of historical modeled storm tides beginning in 1950 can enhance our understanding of US coastal hazard. Our newly developed database of TC sizes and associated storm tides/surges can aid in understanding North Atlantic TC climatology and modeling TC wind, storm surge, and rainfall hazard along the US Atlantic and Gulf coasts.
more »
« less
A Generative Super‐Resolution Model for Enhancing Tropical Cyclone Wind Field Intensity and Resolution
Abstract Extreme winds associated with tropical cyclones (TCs) can cause significant loss of life and economic damage globally, highlighting the need for accurate, high‐resolution modeling and forecasting for wind. However, due to their coarse horizontal resolution, most global climate and weather models suffer from chronic underprediction of TC wind speeds, limiting their use for impact analysis and energy modeling. In this study, we introduce a cascading deep learning framework designed to downscale high‐resolution TC wind fields given low‐resolution data. Our approach maps 85 TC events from ERA5 data (0.25° resolution) to high‐resolution (0.05° resolution) observations at 6‐hr intervals. The initial component is a debiasing neural network designed to model accurate wind speed observations using ERA5 data. The second component employs a generative super‐resolution strategy based on a conditional denoising diffusion probabilistic model (DDPM) to enhance the spatial resolution and to produce ensemble estimates. The model is able to accurately model intensity and produce realistic radial profiles and fine‐scale spatial structures of wind fields, with a percentage mean bias of −3.74% compared to the high‐resolution observations. Our downscaling framework enables the prediction of high‐resolution wind fields using widely available low‐resolution and intensity wind data, allowing for the modeling of past events and the assessment of future TC risks.
more »
« less
- Award ID(s):
- 2019625
- PAR ID:
- 10579841
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Journal of Geophysical Research: Machine Learning and Computation
- Volume:
- 1
- Issue:
- 4
- ISSN:
- 2993-5210
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
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
-
Abstract Mountain winds are the driving force behind snow accumulation patterns in mountainous catchments, making accurate wind fields a prerequisite to accurate simulations of snow depth for ecological or water resource applications. In this study, we examine the effect that wind fields derived from different coarse data sets and downscaling schemes has on simulations of modeled snow depth at resolutions suitable for basin‐scale modeling (>50 m). Simulations are run over the Tuolumne River Basin, CA for the accumulation season of Water Year 2017 using the distributed snow model, SnowModel. We derived wind fields using observations from either sensor networks, the North American Land Data Assimilation System (12.5 km resolution), or High‐Resolution Rapid Refresh (HRRR, 3 km resolution) data, and downscaled using terrain‐based multipliers (MicroMet), a mass‐conserving flow model (WindNinja), or bilinear interpolation. Two wind fields derived from 3 km HRRR data and downscaled with respect to terrain produced snow depth maps that best matched observations of snow depth from airborne LiDAR. We find that modeling these wind fields at 100 and 50 m resolutions do not produce improvements in simulated snow depth when compared to wind fields modeled at a 150 m resolution due to their inability to represent wind dynamics at these scales. For input to distributed snow models at the basin‐scale, we recommend deriving wind fields from high resolution numerical weather prediction model output and downscaling with respect to terrain. Future studies should compare suspension schemes used in blowing snow models and investigate wind downscaling schemes of complexity between statistical and fluid dynamic models.more » « less
-
We propose STSRNet, a joint space-time super-resolution deep learning based model for time-varying vector field data. Our method is designed to reconstruct high temporal resolution (HTR) and high spatial resolution (HSR) vector fields sequence from the corresponding low-resolution key frames. For large scale simulations, only data from a subset of time steps with reduced spatial resolution can be stored for post-hoc analysis. In this paper, we leverage a deep learning model to capture the non-linear complex changes of vector field data with a two-stage architecture: the first stage deforms a pair of low spatial resolution (LSR) key frames forward and backward to generate the intermediate LSR frames, and the second stage performs spatial super-resolution to output the high-resolution sequence. Our method is scalable and can handle different data sets. We demonstrate the effectiveness of our framework with several data sets through quantitative and qualitative evaluations.more » « less
-
Abstract Existing weather prediction products suffer from low spatiotemporal resolution, which fails to meet the requirements for fine-grained power forecasting and intelligent control of wind farms. This limitation adversely impacts wind turbine load management and the operational reliability of power systems. To tackle this, the study explores four-dimensional wind field superresolution, aiming to reconstruct high-resolution wind fields that fulfill both temporal and spatial resolution requirements through deep learning methods. The study emphasizes model architecture innovations to effectively capture the spatiotemporal coupling characteristics of wind field variations within wind farm regions. A novel model, termed 4dST-DMS, is proposed, comprising a spatial module and a temporal module, both built upon a 3D extension of the downsampled skip-connection/multi-scale framework as the backbone. Additionally, an iterative interpolation prediction method is proposed in the temporal module to extract multi-scale temporal information and capture dynamic variations, ensuring high fidelity in the reconstructed wind field. The proposed method is validated using the WTK-LED-5min dataset under experimental settings, including a 10-fold spatial resolution enhancement (both longitude and latitude), a 4-fold height resolution enhancement, and a 12-fold temporal resolution enhancement. The results demonstrate that the 4dST-DMS outperforms conventional interpolation methods and existing deep learning approaches in terms of numerical accuracy, vector field continuity and smoothness, structural similarity to real wind fields, and physical fidelity.more » « less
