Influence of availability of TAO data on NCEP ocean data assimilation systems along the equatorial Pacific

It is reasonable to assume that the quality of an ocean reanalysis will depend not only on the ocean data assimilation system but also on the availability of observational data. Among the data assimilated in the tropical Pacific, observations from the Tropical Atmosphere Ocean (TAO) moored array are an important contributor. Due to various reasons, in the last couple of years, TAO had a very low data delivery rate at some of the mooring sites. In this work, we examined the influence of the availability of in situ TAO data from all 10 equatorial mooring sites in 1999-2014 by comparing ocean temperature from two data assimilation systems at National Centers for Environmental Prediction (NCEP): the Global Ocean Data Assimilation System (GODAS) and the Climate Forecast System Reanalysis (CFSR). Our working hypothesis is that missing of in situ observational data should lead to less observational constraint, and consequently, larger divergence between ocean reanalyses. Compared with periods of no in situ TAO observations, the disagreements of ocean temperature between CFSR and GODAS are indeed smaller when in situ TAO observations are available. The disagreements are largest along the thermocline due to its being the region of strongest variability. Thus, it is suggested that without the constraint of in situ TAO observations, the inconsistency between the ocean reanalyses from two NCEP data assimilation systems is larger, highlighting the contribution of TAO observations. Moreover, it is also suggested that the disagreement of ocean temperature between GODAS and CFSR is mainly caused by the difference of the models used in the assimilation systems and the impact of in situ TAO availability plays a secondary, but important role. To improve the quality of the assimilation products, it is necessary both to reduce model bias and to assimilate more high-quality data.