*2.3. Ocean Surfaces*

The primary goal of the CYGNSS mission is to provide insight into the rapid intensification of TCs and to better measure their windspeeds. To this end, a number of studies since the mission's inception have focused on the application of CYGNSS data products to the study of TCs and tropical convection. Up until the release of the v3 winds data product, most of these studies have used simulated CYGNSS data for algorithm development. For example, Morris and Ruf developed parametric methods for filling in the gaps in the CYGNSS measurements [59], and for characterizing the size, structure, and strength of

a TC [60]. Several studies have used simulated CYGNSS winds to develop assimilation methods for improving tropical cyclone track and intensity forecasts [61–64]. Simulated CYGNSS winds have also been used to demonstrate convective activity monitoring [65]. With the successful development of the v3 winds data products, these studies have transitioned to the application of actual CYGNSS observations for applications such as improving storm center estimates [66] and developing a surface heat flux product [25]. Sessions 5–6 of the Science Team meeting focus on the progress made for ocean-based applications: TCs and other science applications, improving weather forecasts, and ocean altimetry.

Session 5: Altimetry and Tropical Cyclones and Tropical Convection I

A new approach has been developed to incorporate CYGNSS winds into storm surge models, using machine learning to determine a predictive model for CYGNSS winds in the future, based on Global Forecast System (GFS) forecast data and parametric model representation of the CYGNSS inner core TCs winds. The use of CYGNSS in conjunction with ancillary satellite data to measure diurnal wind variations was presented, including also initial results from a "rapid change detector" that is being developed to detect rapid changes to the wind field, which could be used to signal convective activity. Preliminary results of identifying early cyclogenesis from easterly waves using CYGNSS data were highlighted. The modality of the identification comes from the findings of increased L2 latent heat flux and surface wind speeds leading up to tropical cyclone genesis. An update on comparisons of CYGNSS MSSs to those derived from buoy measurements and a coupled atmospheric-wave-ocean model, confirmed the importance of including short wave contributions to the MSS in the models to match the observed MSS values. Validation efforts have been performed for the CYGNSS CDR v.1 products using data from microwave radiometers, including SMAP, WindSat and Advanced Microwave Scanning Radiometer (AMSR-2). A comparison of the wind speeds indicates that the CDR v.1 Young Sea/Limited Fetch (YSLF) winds are more accurate than the v2.1 wind speeds but still are poorly correlated to the other microwave radiometer wind speeds. The CYGNSS-NOAA wind speeds however, correlated well with the SMAP, AMSR2 and WindSat winds. Results of data assimilation of CYGNSS L1 DDMs into ECMWF background winds, where the resulting wind speeds, with and without the CYGNSS L1 DDM, were compared to scatterometer winds from the ASCAT-A, and B and the OceanSat Scatterometer (OSCAT).

The assimilation of the CYGNSS data (Figure 9) improved the ECMWF background at specular points predominantly for wind speeds < 15 m/s. Matched filter retrievals were applied for maximum CYGNSS TCs winds, using the Willoughby–Darling–Rahn model [67]. After eliminating storms that violated the model's assumptions, good results were obtained for the matched filter output. An update of CYGNSS L1 data for ocean altimetry was presented, showing a number of improvements and corrections, including waveform pre-processing, delay compensation, and re-tracking. New results were presented for a case study in the Caribbean. Finally, an overview of several CYGNSS-based altimetry methods was presented, showing an accuracy on the order of several meters. Future improvements are likely to be achieved in accounting better for tides, and ionosphere and troposphere delays.

Session 6: Tropical Cyclones and Tropical Convection II

The final session of the June 2020 CYGNSS Science Team meeting focused on the use of the L2 wind speed products for improving forecasts and wind field analyses of TCs and tropical convection. CYGNSS provides wind speed data that can be used for storm surge predictions, which is a major source of destruction for communities lying in the paths of a storm. Simulations showed the storm surge from the Hurricane Harvey (2017), which were driven in part by wind speed observations from CYGNSS. Efforts were performed to diagnose the structure of TCs' wind fields using the CYGNSS wind speed observations and found that, given good quality data from CYGNSS, the size of the wind field can be determined. A methodology for creating storm wind fields that move with the storm over time was also presented (Figure 10). These datasets are available in PODAAC, alongside other CYGNSS products [68].

**Figure 9.** Assimilating CYGNSS DDMs into ECMWF background winds to improve surface wind forecasts: [**top**] demonstrating the ECMWF background at 150 km resolution, [middle] analysis field with the DDM assimilation, and [**bottom**] the difference between the two. Image by F. Huang et al.

Apart from analysis of storm wind fields, the observations from CYGNSS are also being used to improve weather forecasts. Efforts to improve the initialization of weather model simulations with CYGNSS observations were presented, including also information about where CYGNSS improves the forecasts of TCs. Throughout the session, team members compared techniques for quality control and optimal ingestion of wind speed data from CYGNSS.

CYGNSS is also playing a role in studies of tropical convection, where other observations suffer in heavy precipitation. Simulations of tropical convection with a coupled atmosphere, ocean, wave model were described, along with a comparison of simulated winds with those observed by CYGNSS. A number of on-going investigations examined CYGNSS observations near tropical oceanic thunderstorms, including an examination of the characteristics of the data in storms with and without lightning [69]. The importance of wind-driven fluxes on the Madden Julian Oscillation (MJO), using CYGNSS data was highlighted [70]. Finally, it is worth highlighting that several team members are using CYGNSS data to study the processes of air-sea interaction in weather and climate science [71].

**Figure 10.** CYGNSS-derived wind speed retrievals over the Hurricane Florence by D. Mayers et al.
