A250-05
The Influence of Assimilating CYGNSS DDMs on Global Analyses and Tropical Cyclone Forecasts
Thursday, 17 December 2020: 04:20
Virtual
Feixiong Huang1, Bachir Annane2, James L Garrison1, Mark Leidner3, Giuseppe Grieco4, Ad Stoffelen5 and Ross Hoffman6, (1)Purdue University, School of Aeronautics and Astronautics, West Lafayette, IN, United States, (2)University of Miami, CIMAS, Miami, FL, United States, (3)Atmospheric and Environmental Research, Lexington, MA, United States, (4)Institute of Marine Sciences, Barcelona, Spain, (5)Royal Netherlands Meteorological Institute, De Bilt, Netherlands, (6)Cooperative Institute for Marine and Atmospheric Studies Miami, Miami, FL, United States
Abstract:
The use of direct remote sensing measurements has a lot of advantages over the use of retrievals in data assimilation. The Cyclone Global Navigation Satellite System (CYGNSS) mission launched in 2016 provides abundant surface wind measurements over the global tropics using the GNSS-reflectometry (GNSS-R) technique. In this study, the CYGNSS Level 1 measurements, delay-Doppler maps (DDMs) rather than the Level 2 retrieved wind speeds are assimilated, as the DDM contains information that is not used in retrieving the Level 2 wind speed. The DDM assimilation is demonstrated both in a global case and a regional case.
In the global case, CYGNSS DDMs are assimilated into the ECMWF 10-m surface wind analyses by a Variational Analysis Method (VAM). Results from one month of data are assessed by collocated scatterometer (ASCAT, OSCAT) winds. The root-mean-square error (RMSE) and bias of wind speeds at specular points are reduced from 1.17 to 1.07 m/s and -0.14 to -0.08 m/s, respectively. Wind vectors at specular points from the analyses, the VAM-DDM winds, are shown to have smaller error and bias than other CYGNSS wind products.
In the regional case, the VAM-DDM wind vectors are assimilated into the NOAA's operational HWRF model and GSI data assimilation system in an Observation System Experiment (OSE) for Hurricane Michael in 2018. The errors of intensity, track and minimum central pressure are significantly reduced in the 4-day forecast compared to the control.
