Geolocation, Calibration and Surface Resolution of CYGNSS GNSS-R Land Observations

Abstract

This paper presents the processing algorithms for geolocating and calibration of the Cyclone Global Navigation Satellite System (CYGNSS) level 1 land data products, as well as analysis of the spatial resolution of Global Navigation Satellite System Reflectometry (GNSS-R) coherent reflections. Accurate and robust geolocation and calibration of GNSS-R land observations are necessary first steps that enable subsequent geophysical parameter retrievals. The geolocation algorithm starts with an initial specular point location on the Earth’s surface, predicted by modeling the Earth as a smooth ellipsoid (the WGS84 representation) and using the known transmitting and receiving satellite locations. Information on terrain topography is then compiled from the Shuttle Radar Topography Mission (SRTM) generated Digital Elevation Map (DEM) to generate a grid of local surface points surrounding the initial specular point location. The delay and Doppler values for each point in the local grid are computed with respect to the empirically observed location of the Delay Doppler Map (DDM) signal peak. This is combined with local incident and reflection angles across the surface using SRTM estimated terrain heights. The final geolocation confidence is estimated by assessing the agreement of the three geolocation criteria at the estimated surface specular point on the local grid, including: the delay and Doppler values are in agreement with the CYGNSS observed signal peak and the incident and reflection angles are suitable for specular reflection. The resulting geolocation algorithm is first demonstrated using an example GNSS-R reflection track that passes over a variety of terrain conditions. It is then analyzed using a larger set of CYGNSS data to obtain an assessment of geolocation confidence over a wide range of land surface conditions. Following, an algorithm for calibrating land reflected signals is presented that considers the possibility of both coherent and incoherent scattering from land surfaces. Methods for computing both the bistatic radar cross section (BRCS, for incoherent returns) and the surface reflectivity (for coherent returns) are presented. a flag for classifying returns as coherent or incoherent developed in a related paper is recommended for use in selecting whether the BRCS or reflectivity should be used in further analyses for a specific DDM. Finally, a study of the achievable surface feature detection resolution when coherent reflections occur is performed by examining a series of CYGNSS coherent reflections across an example river. Ancillary information on river widths is compared to the observed CYGNSS coherent observations to evaluate the achievable surface feature detection resolution as a function of the DDM non-coherent integration interval.