Research On HY-2 Scatterometer Data Assimilation

Before the 1980 s, people's access to the sea wind, sea surface height, salinity, tides and other marine elements mainly rely on ships, buoys, aircraft and other means of access.These means is limited to not only low efficiency, but also high cost, and the data obtained are sporadic point data distribution, and in some extreme conditions, such as the north and south poles, high winds and typhoons centers, traditional measurement methods are difficult to implement. In this case, spaceborne scatterometer emerged. Scatterometer also known as active strabismal microwave sounding unit, which is a kind of non-imaging satellite radar sensor. Sea surface roughness information is obtained by measuring the scatterometer backscatter, thus the wind vector can be retrieved. The high resolution,timeliness and coverage of scatterometer observations can effectively compensate for the deficiencies of conventional observations of the sea, which makes it the primary means of ocean surface wind observation, and is of great importance to the numerical forecast of our marine area.In this paper,we mainly focus on the data assimilation process of the scatterometer of HY-2 Satellite, and do research on data pre-processing, wind retrieval, bias correction and the 3DVAR of the scatterometer observation in detail.The pre-processing of the scatterometer observation is the basis of the wind retrieval,This paper has given the pre-processing procedure of the original data of scatterometer, including the geometric positioning of the radar observation and calculation of the backscattering coefficient. Thinning processing is another aspect of the pre-processing. Besides,the observations need to be screened using the quality flag, to detect the effect of sea-ice and land.Since the dual nature between the backward scattering cross section and azimuth,radar echo signal anisotropy, non-linearity of geophysical model function and signal noise,making the wind out of the maximum likelihood estimation(MLE) Inversion is not unique. Therefore, the direct use of the retrival winds in data assimilation needs a process called ambiguity removal. We introduce a ambiguity removal process which called median filter algorithm which mainly based on its own structure of the wind field. Then we put our own ambiguity removal process which is called PERSCAT method, it combines the information of the background field and the possibility of the ambiguity wind and the experience shows that it gives a good ambiguity removal result of the retrival winds. Before the operational use of the winds after ambiguity removal, it need to be revised in a process called bias correction. We use the triple collocation method which combined with the accuracy of the buoy wind and the consistent of NWP wind, make it more applicable to numerical models and assimilation systems.Currently, the wind in most data assimilation systems is not directly assimilated in the form of wind speed and wind direction, but the wind component u,v under the assumption that they are not related. However, u,v components is decomposed by the wind vector(wind speed, wind direction), so they have some relevance, and also subject to the impact of wind direction error. In the actual wind observations, wind speed and direction are generally observed by different means of access, both of which can be considered irrelevant. Therefore, considering assimilating the wind speed and direction directly if of more positive significance than the u, v component. So we design a wind speed and direction operator, and the numerical experiment shows that it gives a better result than the u,v means.