GNSS-R Processing and NWP Assimilation

In the context of the OSI SAF Visiting Scientist Program, Feixiong Huang from the Purdue University, West Lafayette, USA, worked on the assimilation of GNSS-R measurements into numerical weather prediction (NWP) models. This work took place in May 2019 to August 2019 and was supervised by Ad Stoffelen (KNMI) and James Garrison (Purdue).

Introduction

Global Navigation Satellite System Reflectometry (GNSS-R) measurements are obtained by the reflections of Global Navigation Satellite System signals from the ocean surface and received by a dedicated instrument. Both the UK TDS-1 and USA CYGNSS mission pioneer this technique, where KNMI is supported by EUMETSAT to extend its wind scatterometer expertise to GNSS-R through a EUMETSAT fellow. Recent developments on GNSS-R data processing show that Delay-Doppler Map (DDM), the fundamental measurement of GNSS-R can be assimilated into a two-dimensional wind field using a DDM forward model and a variational analysis method (VAM). Although the results are promising, they are still under the influence of the specular point induced distortions and delay-Doppler ambiguity. We propose to validate the feasibility of assimilation of GNSS-R measurements into NWP models, find potential issues and assess their impact.

Objectives and framework of the study

• Assimilate CYGNSS measurements into ECMWF wind field background and validate the result using scatterometer data;

• Assess the impact of the specular position error on the DDM assimilation;

• Assess the impact of the delay-Doppler ambiguity.

Report conclusion

A two-dimensional Variational Analysis Method has been implemented to assimilate CYGNSS DDMs into ECMWF backgrounds for the data of one month in 2017. The analysis has been validated using the collocated scatterometer data. The result shows that DDM assimilation can be beneficial, but is very sensitive to the power estimation error in which the bias in the transmitter power dominates. The speckle noise in the observation can be another source of error and the background check can be used as the quality control.

The impact of the specular position error on DDM assimilation was shown to be trivial as long as the forward model uses the precise specular position and specular bin index from the CYGNSS Level 1 data. The delay-Doppler ambiguity was found more serious in the cases at high incidence angle. The ambiguity distance was defined to quantify it and select ambiguity-free DDM bins. It is more necessary to use ambiguity-free observations in the GNSS-R assimilation into regional/hurricane models than for global NWP models.

Finally, a scheme of the bias correction method has been proposed to reduce the power bias in the DDM. The power bias correction should be conducted before the assimilation of DDMs and will be further investigated in future studies.

Benefits for the SAF

• GNSS-R can provide winds at high temporal frequency for OSI SAF NWP users;

• This research exploits the capability of GNSS-R for NWP applications and identify potential difficulties.

Report on this study: GNSS-R Processing and NWP Assimilation