(5) exp04: data are limited over the Eastern Pacific region,
(6) exp05: add 2N and 2S data to exp02 case
(7) exp06: Only three locations are used (5N,0,5S,165E),
(8) exp07: Only three locations are used( 140E,160E and 180E,0).
Data assimilation period is from March,1980 to December,1983.
Initial state was a climatological value between 1980 and 1990. Here, it should be noted that we assume that surface wind and atmospheric temperature data is available in the dense observational network over the ocean(Fig.2). This is based that surface data is available due to various observational tools, such as especially satellite scattrometer and altimeter.
3.Results
Fig.3 displays that three year average of depth-longitude crosssection of temperature fields along the equator. It should be noted that the main thermocline of the" truth" fields is relatively located upward compared with the model used in the assimilation. This does not indicate that the performance of the high resolution model is better than that of the assimilation model. It is very important that the model climatology is different from each other, because there still exists a large discrepancy between models and the nature and in order to take this situation into account, the difference is very important.
Fig.4 shows that mean difference of exp00 and expOl. In other words, the diffrence between the analysis fields and the truth fieldsCompared with the climatological wind stress, it is clearly noted that the time-dependent wind stress can reduce the error considerably.
Fig.5 (a) shows the RMSE(Root Mean Square Error) in the expo2, where full subsurface temperature data were used. It is noted that the error is reduced considerably. Next, impacts due to the data distribution is considered. If available data is limited in space, there still exists an impact( see, Fig.5(b) and (c)). In our experiments, the data in the Western Pacific has more impacts than in the Eastern Pacific, because errors propagates into the Eastern Pacific region if the data is limited in the Eastern Pacific region.
In Fig.6, the RMSE in the upper 400m averaged over the tropical ocean (between 2N and 2S from 1 40E to 100W) of exp00, expOl, exp02, exp03 and exp04 are displayed. Within a few months, the error is reduced considerably. In general, we can conclude that the data given by the buoy network has a considerable impact on the monitoring of the tropical upper ocean.
Even though no velocity information are not used, much improvement of velocity field is realized by inserting temperature data. In Fig.7, the RMSE of u along the equator of expoo(Fig.7(a)), and exp0l(Fig.7(b)). At the same time, the RMSE of expo2 is shown in Fig.8. It is noted that much improvement is realized by inserting temperature data in the assimilation system.
Finally, we can say that even three buoy data has an impact on analysis (figures not shown).
