Assimilation of temperature and salinity using isotropic and anisotropic recursive filters in Tropic Pacific
- PDF / 935,067 Bytes
- 9 Pages / 595 x 842 pts (A4) Page_size
- 53 Downloads / 163 Views
Assimilation of temperature and salinity using isotropic and anisotropic recursive filters in Tropic Pacific LIU Ye1∗ , ZHAO Yanling2 1 2
National Marine Environmental Forecasting Center, Beijing 100081, China No 61741 Army Troop, Beijing 10081, China
Received 6 March 2009; accepted 27 January 2010 ©The Chinese Society of Oceanography and Springer-Verlag Berlin Heidelberg 2011
Abstract A data assimilation scheme used in the updated Ocean three-dimensional Variational Assimilation System (OVALS), OVALS2, is described. Based on a recursive filter (RF) to estimate the background error covariance (BEC) over a predetermined scale, this new analysis system can be implemented with anisotropic and isotropic BECs. Similarities and differences of these two BEC schemes are briefly discussed and their impacts on the model simulation are also investigated. An idealized experiment demonstrates the ability of the updated analysis system to construct different BECs. Furthermore, a set of three years experiments is implemented by assimilating expendable bathythermograph (XBT) and ARGO data into a Tropical Pacific circulation model. The TAO and WOA01 data are used to validate the assimilation results. The results show that the model simulations are substantially improved by OVALS2. The inter-comparison of isotropic and anisotropic BEC shows that the corresponding temperature and salinity produced by the anisotropic BEC are almost as good as those obtained by the isotropic one. Moreover, the result of anisotropic RF is slightly closer to WOA01 and TAO than that of isotropic RF in some special area (e.g. the cold tongue area in the Tropic Pacific). Key words: recursive filter, anisotropic, isotropic, background error covariance
1 Introduction More and more oceanic observations are available from different observation platforms (e.g., Ships, observation stations, Buoys, Satellites, TAO, Argo). The goal of this investigation is to show how to apply these large volumes of observations to forecast ocean state and its variability using appropriate data assimilation schemes. The typical assimilation scheme assumed that the background error covariance (BEC) is nearly homogeneous and isotropic (Parrish and Derber, 1992; Courtier et al., 1998; Gauthier et al., 1998; Cohn et al., 1998; Lorenc et al., 2000). However, the complicated ocean usually presents inhomogeneous states. To represent the highly inhomogeneous error covariance, many researches focus on advanced assimilation methods, such as the Ensemble Kalman filter (Evensen, 2003, 2004; Zheng and Zhu, 2008; Zheng et al., 2006), Ensemble Kalman Smoother (EnKS) (Evensen and Leeuwen, 2000); the Singular Evolutive Extended Kalman (SEEK) (Pham 2001); Reduced
Rank Square Root filter (Bertino et al., 2002), Recursive filter (Purser et al., 2003a, b) and so on. Advanced analysis systems were developed further to enhance the global or regional ocean simulation. Dobricic and Pinardi(2008) developed an oceanographic variational assimilation method applying RF to the horizontal arithmetic operator of BEC. F
Data Loading...
