A primary result of 3-D variational assimilation of the Argo and TAO Data
International Center for Climate and Environment Sciences (ICCES)
Institute of Atmospheric Physics (IAP.CAS) ,100029 Beijing
Abstract: A new set of gridded data which can better descript the variation of ocean temperature and salinity is achieved by assimilating Argo data and TAO data. The assimilation method used here is 3-D variational which takes the WOA98 as the background.
In the 3-D variational assimilation method, the background error covariance matrix B is assumed to be an error structure function which can be separated in horizontal and vertical direction, that is, the correlation between any two points in the 3-D space can be expressed into the product of horizontal distance function and vertical distance function. A eigenvalue decomposition is performed in vertical direction, the vertical correlation is considered by using typical mode. A Gauss correlation analytic model is adopted in horizontal direction, the calculation of big matrix is realized by recursive filter method. By doing so, the calculation process is greatly simplified, the calculation is reduced and the convergence of the minimization is solved. The method presented here adopted a scale-variable method LBFGS of limited memory to perform the minimization estimation, seek for the minimum of objective function and get the 3-D variational assimilated ocean data which can better represent the characteristics of ocean temperature and salinity variation.
By comparing the temperature field before and after Argo, TAO data assimilation according to the OI SST standard published by NCEP, we can see that the assimilated temperature field can better represent the distribution of SST especially in tropic and subtropic regions. A 30℃ warm center is occurred and the 28℃ isotherm extend to eastern pacific ocean from mid-pacific.