An improved wind retrieval algorithm for the HY-2A scatterometer

Since January 2012, the National Satellite Ocean Application Service has released operational wind products from the HY-2A scatterometer (HY2-SCAT), using the maximum-likelihood estimation (MLE) method with a median filter. However, the quality of the winds retrieved from HY2-SCAT depends on the sub-satellite cross-track location, and poor azimuth separation in the nadir region causes particularly low-quality wind products in this region. However, an improved scheme, i.e., a multiple solution scheme (MSS) with a two-dimensional variational analysis method (2DVAR), has been proposed by the Royal Netherlands Meteorological Institute to overcome such problems. The present study used the MSS in combination with a 2DVAR technique to retrieve wind data from HY2-SCAT observations. The parameter of the empirical probability function, used to indicate the probability of each ambiguous solution being the "true" wind, was estimated based on HY2-SCAT data, and the 2DVAR method used to remove ambiguity in the wind direction. A comparison between MSS and ECMWF winds showed larger deviations at both low wind speeds (below 4 m/s) and high wind speeds (above 17 m/s), whereas the wind direction exhibited lower bias and good stability, even at high wind speeds greater than 24 m/s. The two HY2-SCAT wind data sets, retrieved by the standard MLE and the MSS procedures were compared with buoy observations. The RMS error of wind speed and direction were 1.3 m/s and 17.4A degrees, and 1.3 m/s and 24.0A degrees for the MSS and MLE wind data, respectively, indicating that MSS wind data had better agreement with the buoy data. Furthermore, the distributions of wind fields for a case study of typhoon Soulik were compared, which showed that MSS winds were spatially more consistent and meteorologically better balanced than MLE winds.