 Good afternoon, everyone. I'm Bo Wu. I come from the Institute of Atmospheric Physics, Chinese Academy of Sciences in China. The title of my presentation is Decadal Predictions by Las Jai Pes Capital GCM, FGOS S2. It includes three parts. The first one is about the scale evaluations of the Decadal Predictions by FGOS S2. In the second part, I would like to introduce our new enolization schemes designed for the decadal prediction experiment of the FGOS S2 in Decadal Prediction Project, which is a sub-project of CMAP6. Finally, I would like to introduce our preliminary result about how to predict decadal variabilities of the East Asian San Monsong. Firstly, I would like to introduce the design of enolization schemes. This scheme is used in the decadal predictions by FGOS S2 in CMAP5. We use the incremental analysis update scheme. I use the scheme to assimilate observational oceanic temperature and salinity of up to 1,000 meters derived from graded objective analysis data. UN3, this data is produced by the Met Office Highlight Center. Compared to the conventional lodging scheme, IU scheme can keep the analysis increment constant during the assimilation, so it can effectively suppress the noise in the assimilation. We use anomaly assimilation approach, so no clear drift during the prediction, so we don't conduct any posterior correction. Hindercast and forecast rounds follow the CMAP5 protocol. Because the limitation of computing resources during that time, the hindercast and the forecast has started every five years, and we have three members. This figure shows the hindercast qualities of surface temperature. We use two metrics from the IPCCI5 to measure the predictive skills of FGOS S2. The left panel is for hindercast year 2 to 5, and the right panel is for 6 to 9 years. The first matrix is the IMSSS, is the Rootman-Square Scale Score. The high score, the high predictive scale is. The second matrix is the ratio between Rootman-Square, IMSC of the decodal predictions and IMSC of historical rounds. The only difference between these two experiments is that in the digital prediction experiment, linearization are conducted. So we hope that the linearization can bring some added value for the prediction. So the lower value, the higher predictive skills. For the IMSSS, we can find that FGOS shows significant high predictive skills in the ocean, chopping on West Pacific and Atlantic. There's no significant difference between the hindercast year 2 to 5 and 6 to 9 years. But if we see this matrix, the ratio of metrics, we can find that the added value of decodal prediction is primarily seen in the Atlantic, because only in these regions the values are significant. For the comparison, I also showed the hindercast quantities of surface at temperature predicted by the CMF5 and multimodal ensemble. This figure is from IPCCI-5, Chapter 11. We can find that basically, if we see the IMSS, we can find that many regions have high predictive skill, especially in the ocean, tropical western Pacific and Atlantic. But if we see the ratio between IMSC of the decodal prediction and IMSC of historical round, we can find that the added value improvement is primarily seen in the Atlantic, North Atlantic and South Atlantic. In other regions, the added value is not significant. Because FGOS has to show high skill in the Atlantic, so we further investigate the skills in predicting Atlantic and multimodal variability. We use two metrics, the correlation on the IMSC is consistent with IPCCI-5, Chapter 11. We can find that both metrics, the skills of decodal prediction is much higher than the historical round. The red line is the decodal prediction result. We can find that the highest predictive skill is reached in the year six to nine is somewhat different from the multimodal ensemble result. I think this may be caused by the limitation of ensemble size, I think. We also show the time series for the in the class of year six to nine. We can find that the black line is the observation. We can find that basically the decodal predictions result is consistent with observation, but much higher predictive skills and historical experiments. Many previous studies proposed that the AMV is highly associated with fluctuation of AMOCK. We checked the model performance in predicting the decodal variability of AMOCK. This figure shows the clampotology AMOCK simulated by the FGOS S2. We can find that its central intensity is about 18-thread loop, generally consistent with observation. Now we calculate the lack correlation between AMV and Northward and its transport. The red line and the black line is for AMV lagga, two years and five years. We can find that the lack correlation is very high in the South Atlantic and the North Atlantic to about 14-5 degrees, not degrees. It's generally consistent with the North edge of the AMOCK. To the North of this degree, the correlation decreased dramatically. If we see the simultaneous correlation, we can find that the correlation is quite low in the North Atlantic. This result suggests that the AMV is driven by the fluctuation of Northward and its transport preceding its transport. We further checked the model performance in predicted decodal variation of AMOCK because there is no true observation for AMOCK. We used the assimilation rounds as of the vision. We can find that for both correlation and the AMSE, the decodal prediction, the scale is much higher than is much higher than the historical, especially the Hindacast years 5 to 8 and 6 to 9. If we see the time series of the Hindacast years 6 to 9, we can find that basically the decodal variation of AMOCK simulated by the decodal predictions experiment is consistent with that simulated by the assimilation rounds. The scale is much higher than that from the historical rounds. Then we would like to introduce the new scheme designed for the FGOs in the next generation decodal predictions. This is a schematic diagram. We merged EON-OI and IU scheme together. In one assimilation cycle, we first integrate the model freely. This produces the first gas for the assimilation. Then we use the EON-OI. EON-OI is in some optimal interpretation. This is generally the same with EON-KF, but I don't need so much computing resources. We use the EON-OI scheme to generate, to calculate the analysis increment. Then we return back to the start of the integration and then integrate the first forward. In the process, we introduce the analysis increment through IU schemes. Finally, I would like to introduce how to predict the decodal variability of Eastern Asian Sun Monsoon. Because we can see that at present, the decodal prediction shows low scale in the Eastern Asian Sun Monsoon, but high scale in the AMV. If we know how the AMV influences the Eastern Asian Sun Monsoon, we can at least partly predict the decodal variation of Eastern Asian Sun Monsoon. We apply the AMV-UF analysis to the GJA mean low-level wing of the Eastern Asian Sun Monsoon region. Then we get to dominant mode. Because time limits, we just show the simple result. The second mode is highly correlated with the AMV. Its special pattern is anti-saclon over here. In the China, there is a strong north wing. It indicates the weakening of the Eastern Asian Sun Monsoon. Then we compare the results from the 20-year observation launch to check the robustness of the mode. It's fine that our analysis result is consistent with the observation. Now, we show the SIT anomaly associated with mode. We can find that it's a generator, AMV. If we see the large-scale circulation, we can find that wave trend here, mid-latitude north hemisphere. This wave trend is driven by the AMV. Generally, it has a biotropical structure. If we show the added component of the gel potential, we can find that this wave trend is more clear. The vector is the wave activity flux. We can find that this wave trend is propagating eastward along the jet stream. It has wave number five structure. Then we compare the result with GA-55. We can find that this wave trend is robust in different renaissance data sets. Finally, we make a conclusion based on the 20-year observation. The second mode is characterized by SACRON anomaly extending from north east China to Japan with a part of inter-decadal wave trend. We call it inter-decadal CGT pattern, SACRON global telecommunication pattern. Inter-decadal CGT pattern is associated with the forcing from AMO. Considering that the decadal prediction can expand in CMF-5 showing high-scale in the North Atlantic, the result may be used in the decadal prediction of East Asian salmon song. Thanks.