Research On Data Fusion Of The Multi-Source Wind Field In Northwestern Pacific Ocean

As an important physical parameter of sea and air interface,sea surface wind field is closely related to the vast majority of seawater movement in the ocean.Sea surface wind field is also an important medium air-sea interaction,which regulates the transport of material and energy between the sea and atmosphere.It also plays an emphasis role in the global climate regulation.The conventional observation system of sea surface wind field mainly concludes ships,marine buoys and coastal stations.Because of the limitation of conventional observation methods,high cost and difficulty to meet the continuity in time and space,satellite remote sensing technology provides the possibility for continuous monitoring of sea surface wind field in large area.At present,spaceborne scatterometer and radiometer are the main sensors for observation of sea surface wind field in large area.Scatterometer is an important remote sensing method to obtain sea surface wind field,which could observe sea surface wind speed and wind direction all weather[1].In addition,the full polarimetric microwave radiometer not only has realized the measurement of some conventional ocean physical parameters in passive,but also has the advantage of the large amount of detection information[2].The accuracies of microwave scatterometer and microwave radiation are different for different wind speed range.The combination of the two kinds of sensors has strong complementarity in the study of sea surface wind field data fusion,which overcomes the shortage of single sensor.In the fusion process of different sources from types of remote sensing sensors,there are phenomena of resolution inconsistencies and data loss due to the coverage and data quality.Then interpolation method is needed according to the wind field data to get more comprehensive wind field data with more accurate resolution.There are integral interpolation and local interpolation methods in common.Local interpolation method takes advantage of higher precision and less calculation,so this kind of interpolation method has been used in this paper.Shepard[3]proposed an improved inverse distance interpolation.Its adjacent points are determined by the combination of distance and number of points,which effectively reduces the complexity and computational complexity of the calculation,and also overcomes the instability caused by many conditions caused by the integral interpolation.Then Li Zhenquan[4]defined an interpolation method,inverse distance weighted interpolation involving position shading,to apply to the calculation of rainfall.It has been proved that this method eliminated the phenomenon of isolated circle appearing in inverse distance interpolation.Based on this idea,Shepard interpolation method involving position shading was considered in this paper.Koch[5]mentioned the Barnes interpolation,a kind of objective analysis,and different parameters are discussed in some cases.He has proved that reliable data could be obtained in a simple way.The Barnes interpolation method has been widely used for many years,especially in numerical prediction[6].The interpolation methods above are good for the small range,so they are not entirely applicable for obtaining sea surface wind field data in large area,such as Northwestern Pacific Ocean.Therefore,a new interpolation method combined serval interpolation methods is proposed to apply to the wind field data in Northwestern Pacific Ocean.Achievements of the theses mainly include the following aspects:(1)The fusion method has been applied to Northwestern Pacific Ocean for some wind field products,which are all-inclusive at the height of 10 meters above sea level with a high resolution(0.125°×0.125°)in 2016.Besides,the weight of sample points at different relative positions of interpolated points would be affected by the position shading for the areas with uneven edge data,which effectively reduces the error of results.(2)According to the annual variation of the sea surface wind field data in different directions of Northwestern Pacific Ocean,the average wind speed in summer is slightly lower than the average wind speed in winter,and the occurrence frequency of high wind speed is also higher than that in summer.In the same longitude,the annual average wind speed increases gradually from south to north,and the frequency of the high wind speed appears also.In the same latitude,the annually average wind speed gradually decreased from ocean to coast,and the frequency of strong wind appeared slightly reduced.(3)According to the seasonal variation of the sea surface wind field in the Northwestern Pacific Ocean in 2016,there are obvious monsoon phenomena in this sea area,especially in summer and winter.The average wind speed in winter is the highest,and the average wind speed in summer is the lowest.(4)Compared with the monthly average sea surface wind field measured by WindSat radiometer,the trend of wind field distribution is basically the same.The correlation coefficient of wind velocity data is 0.84,and the mean square root error is 0.54m/s.Therefore,the similarity of fused wind field data and data obtained from RSS data center are higher.Compared with the buoy data,the correlation coefficient of wind velocity data is 0.71,the mean square root error is 1.16m/s.Compared with the ECMWF reanalysis data,the wind direction distribution is the same trend as the fused results,and the wind speed mean square root error is larger,which is 1.91m/s.