The measurement of ocean surface wind speeds in precipitation from satellite micro-wave radiometers is a challenging task. Rain attenuates the signal that is emitted from the ocean surface. Moreover, the rain and wind signals are very similar, which makes it difficult to distinguish wind from rain.
The rain contamination can be mitigated for radiometers that operate simultaneously at C-band and X-band channels, such as WindSat, AMSR-E and AMSR2. The basic principle is to use combinations between C-band and X-band channels that are sensitive to wind speed but relatively insensitive to rain. Based on this principle, we have developed algorithms for retrieving wind speeds in rain from the WindSat and AMSR sensors. These algorithms are statistical regressions and are trained specifically under tropical cyclone conditions. We lay out the steps of the algorithm development, training, and testing. The major source for training the algorithm is provided by wind speeds from the SMAP L-band radiometer, which have been proven to be reliable in strong storms and are not affected by rain.
We show that the WindSat and AMSR tropical cyclone wind algorithms perform well under precipitation where standard passive wind speed retrievals fail. Our assessment is based on comparisons of the SMAP, AMSR and WindSat TC winds with data from the Hurricane Weather Research and Forecasting (HWRF) model, the Sentinel-1 polarimetric Synthetic Aperture Radar (SAR), and with in-situ measurements from the Stepped Frequency Microwave Radiometer (SFMR) that is flown on-board of hurricane penetrating aircrafts.
We will examine the possibility of extending the C/X-band tropical cyclone wind algorithm to X/K-band channels. Doing so could allow accurate tropical cyclone wind measurements to be taken by the GMI sensor and the upcoming Weather System Follow-On Microwave (WSF-M) mission, which will launch in 2023.
Additionally, an exciting future satellite mission is the European Copernicus Imaging Microwave Radiometer (CIMR), which is currently being developed and is planned to begin operation by the end of this decade. CIMR will take simultaneous observations at all 3 frequencies: L-, C-, and X-band. In addition, the CIMR will have polarimetric channels (3rd and 4th Stokes parameters), which will enable it to measure not only scalar wind speed but also wind vectors, similar to WindSat, but with a wider swath and at better spatial resolution. This will make it an excellent tool for observing tropical cy-clones. We will discuss this mission and give an outlook for extending the current TC wind algorithms to the CIMR sensor.

Today, only the resulting effects of a tropical cyclone passage over ocean can be analyzed with satellite measurements. In particular, there is no remote sensing satellite observing system to inform about horizontal and vertical interior motions within the most intense inner-core area. Harmony’s all weather and high-resolution capabilities will result in the first system capable of providing co-located surface wind, wave and current directional information, at sufficient resolution to resolve both atmospheric and ocean boundary layer characteristics.

By analyzing surface roughness directional variations down to O(500m) scales with different viewing angles, Harmony will have the unique capability to help for the disentanglement of different contributions from the Organized Large Eddies (OLEs) in the tropical cyclone boundary layer and the ocean surface wind direction. This will allow the quantification of near-surface inflow and associated storm boundary layer structure.

We will also discuss known limitations of the single viewing angle of existing SAR missions to estimate directional waves properties. This is particularly true for Tropical Cyclones for which the imaging mechanism is strongly and non-linearly impacted by the extreme wind and waves. The extended Harmony angular probing diversity should significantly help improving the retrieval strategy. This will lead to obtain unique information on the sea-state development within the TC inner and outer core regions.

This work will review the state of the art of C-Band SAR sensors for geophysical parameters extraction in the case of TC. In particular, we will focus on ocean surface wind speed and direction, Organized Large Eddies as well as TC-induced waves. Then we will discuss the benefit of having a bi-static system to better constrain the wind and waves inverse problem. Sentinel-1 data acquired during the Satellite Hurricane Experiment Campaigns will be used in combination with Sentinel-1 data acquired in Wave Mode outside of the waves generation area to provide information on TC waves direction and wavelength.

The center location and the intensity of a Tropical cyclone (TC) are of common concern to both physical oceanographers and weather forecasters, since they are of great significance for improving the accuracy of TC forecast, and in turn reducing the destructiveness of TC events. Satellite scatterometers are able to obtain accurate stream line features in the presence of TCs, as widely used in TC monitoring, notably for TC center location estimation. However, the scatterometer-derived extreme winds are usually underestimated with respect to in-situ GPS dropsonde winds, due to rain contamination, signal saturation, and lack of proper extreme-wind base calibration, therefore limiting the application of its data in determining the TC intensity.

In this paper, the characteristics of scatterometer wind vectors, as well as their divergence and curl, are analyzed. Following the unique pattern of wind stress divergence and curl near the TC core, a new method is developed to determine the TC center location. That is, two positive local maxima and two negative local minima appear symmetrically near the TC core, due to surface convergence around a cyclone, which in turn, creating an area of high pressure aloft and causing air to diverge at upper levels. As such one can take the intersection of the two lines constructed separately by the local maxima and the local minima as the TC center. This technique is applied to the 32 HSCAT and 9 ASCAT acquisitions of TCs over the Western Pacific in 2019. The mean difference between the identified HSCAT/ASCAT TC center and the interpolated best-track positions is about one wind vector cell (~25 km). Moreover, closely collocated (in time) NOAA P-3 Stepped Frequency Microwave Radiometer (SFMR) wind data show a good correspondence between the SFMR minimum wind speed pattern inside the eyewall and the TC centre location as depicted by the new method.

Then the radial extent of 17-m/s winds (i.e., R17) is calculated from the scatterometer wind data. The feasibility of scatterometer wind radii in determining TC intensity is evaluated using the maximum sustained wind speed (MSW) in the China Meteorological Administration best-track database, with the objective of predicting the TC intensity using the scatterometer wind radii information. It proves that the estimated R17 value is better than the maximum wind speed of Ku-band scatterometer in terms of characterizing the TC intensity. However, the correlation between R17 and MSW is still relatively low (r = 0.54) to develop a universal TC intensity prediction model. Through case-by-case analysis, we find that the R17 value is highly correlated with the best-track MSW for each single TC event, implying that the scatterometer wind radii are useful in estimating TC intensity by limiting the concerned spatial region and temporal duration.

In summary, in terms of nowcasting or short-range forecasting purpose, the scatterometer R17 value is quite useful in assessing the evolution of TC intensity. With the joint observations from the current virtual scatterometer constellation, e.g., the HSCATs onboard HY-2 satellites, the ASCATs onboard Metop satellites, and the China-France Oceanography Satellite scatterometer, it may be feasible to monitor the TC evolution in near-real time.

Response of GNSS reflected signals to wave spectra across cyclones:

The signals transmitted by the Global Navigation Satellite Systems (GNSS) can be used for other applications beyond navigation and positioning. Earth remote sensing is one of the opportunistic applications of the GNSS, for example, after the signals bounce off the Earth surface as a bi-static radar (GNSS reflectometry, GNSS-R).

The scattering of the L-band GNSS signals off the sea surface is a diffuse process where the incident power is redistributed towards a broad range of directions. According to this process, the rougher the surface the broader the power distribution and the lower the peak power of the reflected ‘echo’. This principle is the base of the ocean wind measurements with GNSS-R missions such as UK TDS1 and NASA CyGNSS. The former demonstrated the concept from spaceborne receivers with a relatively large set of data, whereas the latter is collecting GNSS-R observables from a 8 low earth orbiters since 2016, with a target on tropical cyclones. The wind speed retrievals obtained from these missions are in good agreement with other spaceborne sensors and numerical weather prediction (NWP) models in the low to moderate wind regime [e.g., 1], while the high wind regime introduces some degree of controversy in the community, with distinctive geophysical model functions for different wind regimes [e.g., 2] and reduced sensitivity at the high end. Other aspects to consider are the swell and the stage of development of the wind-driven waves: L-band signals are sensitive to sea wavelengths longer than approximately half a meter, so there is not sensitivity to the fine surface ripple instantly induced as the wind blows. Some of these aspects are particularly relevant in the sea surface characterization across tropical cyclones and other extreme ocean phenomena.

A way to tackle the high-wind regimes consistently with the low and moderate ones was suggested in [3], where a variational approach was taken to both calibrate the GNSS-R observables and estimate the wind speed correction with respect to background information obtained from NWP models. The approach, conceptually similar to that of data assimilation into NWP models, resulted in post-fit residuals similar to those obtained with co-located measurements by the SMAP radiometer, and slightly lower winds than those of SMOS, in both cases for the range of wind speeds up to 50 m/s.

In this study, funded by the ESA CN 4000132954/20/I-NB, we suggest to apply the initial steps of the variational approach onto wave spectra from oceanographic models rather than wind speed from NWP. The main difference is the richer parametrization and complexity of surface roughness phenomena that can be accounted in wave spectra, compared to wind fields from NWP models. The initial goal is to identify whether the GNSS-R observables across hurricanes and typhoons can distinguish between different wave spectra conditions that might correspond to the same wind speed, but typically occupy different sections of the cyclones. The ensemble of model spectra follows the parametric model developed by Kudryavtsev et al. (2021) [4].

Two-dimensional fields of L-band filtered mean square slopes (mss) are then extracted across cyclones, and fed into the variational system. Its forward operator generates synthetic GNSS-R like observables and their analysis might indicate sensitivity to the spectra variations around the cyclone. Actual CyGNSS data that took measurements along the same cyclonic structures will be used to confirm the sensitivity, and variational retrievals to L-band mss will be attempted.


[1] Foti t al., 2015, doi:10.1002/2015GL064204
[2] Ruf et al., 2019, doi:10.1109/JSTARS.2018.2833075
[3] Cardellach et al., 2020, doi:10.3390/rs12233930.
[4] Kudryavtsev et al., 2021, doi:10.1029/2020JC016915

Satellites have been used to analyse and monitor Tropical Cyclones (TCs) since the 1960s. Today it remains one of the most valuable source of information about TCs. In particular, active scatterometers and passive radiometers onboard polar-orbiting satellites can provide relatively large-scale pictures of TC surface wind fields. They are operationally used by TC centres to characterize TCs.

However, these observations can have very different resolutions in both space and time. For instance,
L-band passive radiometer embedded on ESA’s Soil Moisture and Ocean Salinity (SMOS) mission has a ~40 km resolution, which critically hampers its ability to depict the wind structure inside the TC eye (~10 – 100 km diameter). This makes the assessment of the high wind speed gradients/areas in the TC eyewall very difficult. In particular, this sensor poorly estimates the Radius of Maximum Winds (RMW), an important parameter of the TC structure. Wind products from scatterometry such as the ASCAT sensors embedded on MetOp-A, B and C missions have a resolution of ~12.5 – 25 km. They also lack spatial resolution to accurately capture the inner-core TC wind structure and suffer from a loss of sensitivity for the most intense wind speeds.

Today, with Synthetic Aperture Radar (SAR), it is possible to observe the ocean surface within TC at very high resolution (~3 – 100 m) and derive estimates of wind field at ~1 – 3 km spatial resolution. These observations, such as the ones provided by ESA’s Sentinel-1A and B missions, though limited in number, provide very precious information of TCs on both wind speed and directions, including the inner- and outer-cores.

This study aims at combining satellite acquisitions of TCs at different spatial resolutions and from different sensors/missions to produce estimates of 2-dimensional high spatial resolution wind fields. The ability to reconstruct an accurate inner-core wind structure from low spatial resolution data using statistical methods will be assessed. In particular, we will discuss whether it is possible to reconstruct inner-core important parameters such as the RMW using only outer-core wind field information.