Abstract Accurate retrieval of significant wave height (SWH) under tropical cyclone (TC) conditions is essential for marine hazard warning and ocean dynamics studies. Although spaceborne synthetic aperture radar (SAR) offers unique capabilities for observing extreme sea states, existing empirical wave retrieval models often do not fully account for TC wind context. This study presents CWAVE_S1TC, a wind-context-based extension of the CWAVE model, which incorporates three key proxies of TC wind context derived from SAR wind retrievals: the local surface wind speed, the maximum wind speed, and the normalized radial distance, to enhance SWH retrieval. A dataset comprising 63 Sentinel-1 SAR scenes from 40 TC events, together with collocated reference SWH from the ECMWF ERA5 reanalysis and WAVEWATCH III hindcast data, is used for model development and validation. The three proxies show significant correlations with SWH, with correlation coefficients of 0.84, 0.56, and −0.49, respectively. CWAVE_S1TC achieves a coefficient of determination of 0.81 and a root mean square error (RMSE) of 1.03 m, outperforming the basic CWAVE_S1 model in the model-development dataset. Independent validation against WW3 data shows that CWAVE_S1TC reduces the Bias from 0.61 m to 0.37 m and the RMSE from 1.86 m to 1.14 m, while increasing the correlation coefficient from 0.77 to 0.85. Leave-one-storm-out and cross-basin validation further demonstrate model robustness and applicability beyond the model-development region. These results highlight the value of incorporating TC wind-context information into SAR-based SWH retrieval and demonstrate the potential of CWAVE_S1TC for improving SWH retrieval under TC conditions.
Abstract Coastal observing platforms typically lack the spatial resolution needed to monitor the transformation of surface gravity waves, which impact nearshore circulation patterns, wave-driven runup, and sediment transport. Previous studies have demonstrated the effectiveness of deriving surface wave information from X-band radar by applying three-dimensional Fast Fourier Transforms (3D FFTs) to backscatter images in the open ocean. Near the coast, however, X-band retrievals are more challenging to interpret since wave fields are spatially inhomogeneous due to bathymetrically driven shoaling and refraction. Here, the classical four-point slope array method, traditionally applied to pressure sensors, is adapted to radar backscatter imagery for estimating wave direction at meter-scale resolution. Using data collected at a planar beach near Point Sal, CA, slope-array-derived wave statistics are compared with in situ Spotter buoys, a spectral refraction model, and the 3D FFT retrieval method. Additionally, a signal-to-noise-based quality control approach is proposed for coastal radar wave measurements to identify sampling conditions that degrade observations and the optimal wave and meteorological conditions for radar-based wave retrievals at this site are characterized. Quality controlled slope array and 3D FFT-based wave retrievals yield RMSE values of 13° and 11° relative to an in situ Spotter buoy time series. Spatial analyses indicate that slope-array-derived wave directions are in closer agreement with spectral refraction model predictions while providing directional estimates at 12 m spatial resolution, compared to 543 m using the 3D FFT approach. The results demonstrate the ability of X-band radar to capture small-scale coastal wave refraction, providing new opportunities to study nearshore wave dynamics.
Accurate radiometric calibration is crucial for various applications of ocean color satellite sensors. However, tracking the radiometric performance of sensors after satellites enter orbit using ground-based sources is challenging, as is effectively evaluating the accuracy of onboard calibration systems. This study attempted to use the Ocean Color component of the Aerosol Robotic Network (AERONET-OC) data to perform vicarious calibration (VC) of ocean color satellite sensors. Through rigorous analysis of in situ AERONET-OC data, several field measurements that meet the predefined stability thresholds were identified as having the potential to be used for VC. Then, the field multispectral data are converted into hyperspectral data by machine learning models and employed for VC of the Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua and the Chinese Ocean Color and Temperature Scanner (COCTS)-Hai Yang 1D (HY1D). These results were compared with their respective onboard calibration. For MODIS-Aqua, the results were grouped into separate 6-month cycles and showed an average of 38 valid matchups for different AERONET-OC sites over six cycles from 2020 to 2022. Additionally, taking 412 nm as an example, the top-of-atmosphere (TOA) radiances estimated using AERONET-OC measurements fell within the range of 4-12 mW cm-2 mm-1 sr-1, which shows a wide overlap with the range for global waters (3.5-14 mW cm-2 mm-1 sr-1). The findings confirm that VC of ocean color satellite sensors, when conducted using data from multiple AERONET-OC sites across different regions, yields consistent results and enables accurate characterization of sensor performance across a wide dynamic range. This offers a reliable basis for assessing sensor performance in the early stages of launch.
Enhanced measurements of vertically resolved meteorological parameters are required to improve now-casting of high-impact weather (HIW) and to evaluate and enhance the performance of numerical weather prediction (NWP) systems. This is particularly true in urban areas where there exists a lack of upper-air observations. To address this need, Environment and Climate Change Canada (ECCC) established the Southern Ontario Lidar (SOLID) mesoscale network (Mesonet) in and around Toronto. Eight Doppler lidars and one water vapor differential absorption lidar (DIAL) provide near-real-time observations of aerosol backscatter, cloud and planetary boundary layer (PBL) height, and vertical wind and water vapor profiles to aviation and operational weather forecasters. Combined with the neighboring New York State Mesonet, the two Mesonets provide some of the only upper-air observations in this urban area inhabited by more than 26 million people. As part of an ongoing impact analysis, this project will assess the potential of these profiling instruments to improve nowcasting operations and data assimilation efforts. Three case studies demonstrate the applications of the Mesonet's observations, particularly for aviation nowcasting and HIW events, including passing cold fronts that result in sudden and extreme wind shifts (-1808 within 10 min) and the Delta flight 4819 plane crash investigation at Toronto's international airport. Such real-time profiling of upper-air winds and water vapor are shown to aid forecasters and improve their ability to make storm-track adjustments with confidence, provide near-real-time analysis for automated nowcasting systems, and enable in-depth verification of NWP forecasts. SIGNIFICANCE STATEMENT: Providing near-real-time observations of vertical wind profiles in an urban environment is crucial to improve nowcasting, particularly during high-impact weather (HIW). The Southern Ontario Lidar (SOLID) mesoscale network (Mesonet) provides such observations at eight sites, covering several cities in Canada including London, Toronto, and Ottawa, Ontario. Combined with the neighboring New York State Mesonet, this is one of the only sources of upper-air wind observations for a region inhabited by over 26 million people. Case studies of the impact of SOLID measurements are shown for HIW events (passing cold fronts) and for the Delta flight 4819 plane crash at the Toronto airport. SOLID is shown to have a positive impact on nowcasting and aviation operations, a forecaster's situational awareness, and for verification of weather models.
The generation of near-inertial oscillations is difficult to predict in ocean modeling due to their sensitivity to the timing of wind pattern changes and other atmospheric properties. If we can accurately predict the phase of near-inertial oscillations, the resulting daily ocean forecasts can inform a reinforcement learning agent within an unmanned underwater vehicle (UUV) to complete a mission while optimizing energy consumption. This study examines how changing the timing of different atmospheric forcing variables impacts the ability of ocean ensemble runs to capture uncertainty in processes triggering near-inertial motions in the Mississippi Bight (MSB) region. The atmospheric forcing does not resolve land-sea breeze variability, limiting the representation of the diurnal variability in this region. Model results are evaluated using high-frequency radar and mooring station current data, comparing the phases of the observed and modeled near-inertial motions in the MSB. Isolating the near-inertial signal in the MSB current field is difficult because the diurnal and inertial frequencies are equal at 30 degrees N. We first perform a harmonic analysis to remove tidal components and then apply a bandpass filter to isolate the near-inertial signal. The ensemble range spread evaluates how well each ensemble run captures uncertainty in the near-inertial phase. The cosine similarity of each ensemble member determines which forecast is most representative of the true ocean state and therefore most suitable for informing UUV path planning. We determine that time shifting all atmospheric variables, as opposed to wind speed only or wind speed and atmospheric pressure, generates the most representative ensemble simulation.
