Abstract
This study presents a new retrieval approach for obtaining wind speeds from CyGNSS level-1 observables. Unlike other existing approaches, (1) this one is a variational technique that is based on a physical forward model, (2) it uses uncalibrated bin raw counts observables, (3) the geophysical information content comes from only one pixel of the broader delay-Doppler map, finest achievable resolution in level-1 products over the sea, and (4) calibrates them against track-wise polynomial adjustments to a background numerical weather prediction model. Through comparisons with the background model, other spaceborne sensors (SMAP, SMOS, ASCAT-A/B), and CyGNSS wind retrievals by other organizations, the study shows that this approach has skills to infer wind speeds, including hurricane force winds. For example, the Pearson’s correlation coefficient between these CyGNSS retrievals and ERA5 is 0.884, 0.832 with NOAA CyGNSS results, and 0.831 with respect to SMAP co-located measurements. Furthermore, the variational retrieval algorithm is a simplified version of the more general equations that are used in data assimilation, and the calibration scheme could also be integrated in the assimilation process. Therefore, this approach is also a good tool for analyzing the potential performance of ingesting uncalibrated level-1 single-pixel observables into NWP.
Highlights
CyGNSS is a NASA constellation of satellites with the objective of measuring ocean winds across tropical cyclones [1]
This study presented a variational retrieval algorithm in order to extract wind speed estimates from CyGNSS uncalibrated observables
The uniqueness of this approach can be summarized in the following points:. It uses uncalibrated observables obtained from a single pixel of the signal-to-noise delay Doppler Maps (DDM), of slightly finer spatial resolution than the combination of pixels used in other CyGNSS retrieval approaches
Summary
CyGNSS is a NASA constellation of satellites with the objective of measuring ocean winds across tropical cyclones [1]. Discussions are taking place about the large uncertainties for wind speeds above 20 m/s in CyGNSS data. In order to overcome this issue, the CyGNSS level-2 (L2) wind speed products are generated while using two separate Geophysical Model Functions (GMF), corresponding to well developed and young sea/limited fetch regimes [26]. The retrievals that are based on young seas are used over extreme events, yielding wind speed root mean square (RMS) uncertainties reported at 17% level [27], or 11.3% [28]. There is no continuity between both regimes, which poses an ancillary uncertainty in the range of transition winds
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
