| Sea surface current and wind fields not only play an important role in the marine dy-namic environment,but also are closely related to human activities.According to the chara c-teristics of coastal region ocean current observations and satellite observations of sea surface wind fields,this paper developed a two-dimensional variational data blending method,and used the developed method for the blending of:1)high-frequency coastal radar radial current and ocean current vector at the Taiwan Strait 2)Indian Ocean satellite wind speed observa-tions and wind vector field.The blended results are validated against independent observation or using experiments by separating data into experimental and testing data sets.The specific research contents are as follows:(1)In the study of sea surface current field,this paper develops a Two-Dimensional VARiational Method(2DVAR),which blends radial current observation with surface ocean current vector from the Regional Ocean Model System(ROMS)at the Taiwan Strait to obtain a blended product.The results show that the averaged relative error between the blended product and the radar radial current is reduced from 9.70%to 1.54%compared to the aver-aged relative error between the model output and the radar radial current.In order to further examine the blending method,two independent sampling experiments are designed using the same data mentioned above.In the first experiment,the radar radial flow data is evenly di-vided into blending samples and independent samples in space.The blending samples of radi-al velocity observations are then blended with the model output.The root mean square error(RMSE)of the blended result is 0.19 m/s compared to 0.41 m/s,the RMSE of ROMS model output.In the second experiment,both the radar radial flow and the model output are divided into blending samples and independent samples with a 1:1 ratio,and the blending samples are generated using samples form the two different sources of data.Then the Triple Collocation(TC)method is used to estimate the error variances of the model output,radar observation and blended results.The analysis shows that the averaged error variances of these three data sets are:0.07 m~2/s~2?0.06 m~2/s~2and 0.05 m~2/s~2,which means the blended result has the lowest error among the three data sets.And the error variance of the blended result is significantly reduced at the region further away from the shore compared with the error variance of the independent radar radial flow as well as the error variance of the model output.All the above results show that the blended product is closer to the true current.(2)In the study of sea surface wind field blending,this paper uses an observational oper-ator that can directly use wind speed.Using the ERA reanalysis data as the background vector field in the Indian Ocean region,and the wind speed in CYGNSS,SSM/I,SMAP,ASCAT and Wind Sat satellite remote sensing as observed wind speed,the 2DVAR algorithm is used to blend the above data to generate a multi-source wind field blended product.The results show that,compared with the ERA data,the relative error in wind speed between the two ind e-pendent buoy data and the blended data is reduced by an average of 8.12%.Compared with the background ERA data,the RMSE of the wind field blended data is reduced by 0.19 m/s in the longitude direction,0.07 m/s in the latitude direction,and the wind speed is reduced by0.35 m/s;the correlation coefficient of the wind field bended data is improved,while its av-erage deviation is reduced.In addition,the TC method is used to estimate the error variance of the CCMP reanalysis data,FNL reanalysis and the blended data in the study area.The re-sults show that the average error variance of the blended data is reduced by 2.13 m~2/s~2 com-pared to the CCMP data and 5.73 m~2/s~2 compared to the FNL data.The blended data is the closest to the true wind speed among the three data sets.The analysis results have shown the effectiveness of the 2DVAR method in blending multi-source sea surface wind fields. |
PDF Full Text Request