Wave-effect considering two-step wind speed retrieving algorithm for new GNSS-R satellite, TRITON

Global navigation satellite system reflectometry (GNSS-R) is designed to get the quasi-specular reflection of the GNSS signal over the Earth's surface. The reflected signal of GNSS-R recorded on a Delay-Doppler Map (DDM) may then be used to retrieve wind speed, ocean surface roughness, and latent heat flux over the open ocean and to retrieve soil moisture over the land. The L-band signal from GNSS is transmitted via forward (quasi-specular) scattering geometry, obeying the geometric optics (GO) limit of the Kirchhoff approximation (KA). One of the factors on DDM is bistatic radar cross section (BRCS) which represents sea surface roughness. The ocean surface slope and roughness spectrum sensed by GNSS-R response to the energy transferred to the ocean makes it possible for the wind speed to retrieve from the GNSS-R DDM. L-band (~1.5 GHz) microwave is less sensitive to the rain than higher frequency band signals such as Ku- band and C- band. There is a potential to use the L-band signal to retrieve wind speed over not only the fully-developed sea but over limited-fetch sea under more extreme weather systems, such as under tropical cyclones.

In this study, we will present a wave-considering retrieving wind speed algorithm for the new GNSS-R satellite, TRITON. TRITON (Wind-Hunter Satellite) is designed and manufactured by Taiwan Space Agency (TASA) and will be launched in the first season of 2023.

The wind speed retrieving algorithm contains self-built Level 1b (L1b) and Level 2 (L2) algorithms. The L1b algorithm starts from Level 1a output, power signal DDM in watts. We will introduce the procedure to calibrate the DDM to normalized bistatic radar cross section (NBRCS) in meters and compute the DDM observables (DDMA and LES) for the next level. The performance assessment of the self-developed Level 1 algorithm. Good agreements have been found compared to the CYGNSS results. The correlation coefficient among ~0.5 million DDMA_cygnss and DDMA_Triton points regression is 0.95. The root-mean-squared error is 4.99, with data ranging from 0 to 200, and the scattering index is 0.19.

In the L2 algorithm, wind speed will be retrieved in two steps. In the first step, NBRCS will be used to compute the mean square slope (mss) with the help of the Fresnel reflection coefficient. In the second step, the relationship between wave age, mss, and wind speed developed based on the state-of-art microwave remote sensing study will be applied to retrieve wind speed. Uncertainty of the two-step algorithm will be assessed and compared with the results from the existing one-step geophysical model function algorithm to check the improvement. Results under the fully-developed sea and young-sea limited-fetch condition will be presented.

Keywords: TRITON, wind speed retrieval, calibration, two-step algorithm, wave age, ocean surface roughness