 Good morning to all of you, myself Manish Kumar Joshi and presently I am working as post-doctoral researcher at Research Center for Climate Sciences, Busan National University, Busan Korea. Well, first of all, I would like to thanks Dr. Fred Kucharski and other organizing committee members like Professor David Strauss, Kristiana, and Dr. Andrew for inviting me to attend this advanced school on topical, extra-topical interactions on interstitial time scales. Well, today I am going to talk on the fidelity of state-of-the-art models, climate models in simulating the observed teleconnection between the tropical, South Atlantic, and Ismar. Well, this work is done in collaboration with Dr. Fred Kucharski. And in fact, he is the lead author of this work. And I am very pleased to have this opportunity to present this work in front of you all. Well, this work has been recently published in QJRMS and the outline of my talk is as follows. First of all, I will give a short, first of all, I will discuss the main objectives of the present study, followed by the methodology. Then I will discuss some of the important results of the present study. And finally, I will discuss in short, the main outcomes of the present study. Well, the objective of the present study is to evaluate the fidelity of 32 CME-5 models in simulating the tropical, South Atlantic, Ismar teleconnection. This study addresses the following issues. Are CME-5 models under consideration capable of simulating the tropical, South Atlantic, SSG variability? Do CME-5 models have capability to reproduce the tropical, South Atlantic, and Ismar teleconnection? Are CME-5 models capable of reproducing the atmospheric circulation and convergence, divergent patterns associated with tropical, South Atlantic? And what are the key features that characterize the teleconnection in good compared to the poor models as measured by the tropical, South Atlantic, and Ismar teleconnection? Well, this table shows the list of 32 CME-5 models under consideration along with their modeling groups and resolution. Well, to examine the tropical, South Atlantic, Ismar teleconnection, the tropical South Atlantic index, which is defined as the area average sea surface temperature anomalies over 30 degrees west to 10 degrees and 20 degrees south to the equator is used. Now, because of the possible co-variability with ANSO, the linear relationship between the two, which is obtained by regressing the tropical, South Atlantic index onto the NENO-3.4 index, has been removed from tropical, South Atlantic. Well, the resulting time series is linearly independent of NENO-3.4 index. That is, there is no correlation between the two time series, and therefore, it can be used as ANSO independent index that could influence the Indian muscle. I may remove this. Just slow down. First of all, we will see how the precipitation is simulated in CME-5 models using two Taylor diagrams based on the fidelity in stimulating the annual cycle as well as this representing the climatological monthly mean or precipitation area average over monsoon co-reason and the spatial pattern of climatological seasonal mean precipitation over Indian monsoon region. Well, the Taylor diagram of annual cycle depicts that except these two models, all models are showing correlation greater than 0.9, but some either underestimate or overestimate the variance. In terms of both magnitude and trace, the models which are lying close to this observation line simulate the best annual cycle as compared to other models. On the other hand, the Taylor diagram of spatial pattern of climatological seasonal mean depicts that most of the models show correlation, good correlation, but some either underestimate or overestimate the variance. Now, in order to identify the models that reasonably simulate the climatological seasonal mean precipitation over Indian monsoon region, we defined a criteria that the correlation should be greater than 0.6 and the normalized standard deviation should lie between 0.75 and 1.25. Well, based on this criteria, these two models have the lowest correlation while these models are highly overestimating the climatological seasonal mean precipitation. Well, it is to be noted that the models are not selected based on the above criteria for the analysis of tropical South Atlantic ISMAR teleconnection because all models are to the some extent able to reproduce the ISMAR seasonal cycle and climatology and also a selection would involve the risk of excluding models from further analysis because of ad hoc and subjectively chosen thresholds. Now, we will see how the tropical South Atlantic pattern is simulated in CME5 models. For this, the JJS sea surface temperature anomalies are regressed onto the standardized low-pass filter, standardized tropical South Atlantic index for the observation and 32 CME5 models. Well, most of the models show well-defined spatial pattern of tropical South Atlantic, localized over the South tropical Atlantic and having magnitudes quite comparable with the observed ones. Well, out of 32 CME5 models under consideration, more than 65% of models also show the negative anomalies over the South Atlantic region and this as a combined can be referred to as South Atlantic ocean dipole which will be discussed in the further slides. This figure shows the Taylor diagram of spatial tropical South Atlantic SST regression pattern over the Atlantic domain and this Taylor diagram also shows that all the models show good correlation that is greater than 0.7, but some either underestimate or overestimate the variance. Well, on close observation, it can be seen that these three models are highly overestimating the variance while these five models are highly underestimating it. But overall, the SST regression patterns and its Taylor diagrams reveal that the CME5 models are capable of stimulating the tropical South Atlantic pattern. Now, we will see how the tropical South Atlantic ISMA teleconnection has been reproduced in the CME5 models. Well, for this, the JJS precipitation anomalies are reversed onto the standardized tropical South Atlantic index for observation and 32 CME5 models. Well, the observed precipitation shows, observed precipitation shows, observed precipitation shows the negative anomalies over most parts of India, especially over the Central Indian region and the positive anomalies over the Eastern India which implies that the warm sea surface temperature anomalies over tropical South Atlantic is associated with the reduction of rainfall over Central India while it causes enhancement of rainfall over Eastern India. Well, on the other hand, the models show a diverse behavior. For example, some of the models show the similar regression pattern as seen in observations which are shown by the tick sign while some show the opposite pattern which are shown by the cross sign. While there are also some models which are showing the weak responses which are shown with no sign. Well, consistent with the observations, the multimodal ensemble which has been obtained by averaging across all models also show a dipolar pattern that is the negative anomalies over all India and positive anomalies over Eastern India but the signal is very weak because it has been averaged across good and poor models. Well, this shows the area average of tropical South Atlantic residual precipitation regression maps shown in the previous slide over Indian land points excluding North East region. Well, based on the sign of average regression coefficients, the models are categorized into two groups, good having the negative average regression coefficients and poor having the positive average regression coefficients. Well, this figure shows the ensemble mean of tropical South Atlantic precipitation regressions patterns for 17 good and 15 poor semifinal models. Well, the ensemble mean of good models also closely resembles to the observation as they are showing the negative anomalies over most parts of all India and positive anomalies over the Eastern India which implies that the, during the warm tropical, warm SSG is over the tropical Atlantic, most parts of India will receive below normal rainfall while Eastern India will have above normal rainfall. Well, the ensemble mean of poor models shows just an opposite pattern. Well, it is to be noted that the classification of models into good and poor categories is solely based on the metrics of tropical South Atlantic isomer regressions and it is not related with the overall model performance. Now, in order to get further insight of the crucial elements in the model tropical South Atlantic pattern, the ensemble mean of tropical South Atlantic residual SSG regressions for good and poor semifinal models are constructed and it can be seen that consistent with the observations, both models show a dipolar pattern over the South Atlantic that is warming over the tropical South Atlantic and the equatorial region and cooling over the Southern South tropical Atlantic which has been referred as South Atlantic Ocean Dipole. Well, it can also be seen that the negative pole of this dipole is likely more pronounced and closer to the observations in good models as compared to poor models. Looking at this tropical South Atlantic SSG pattern, it is difficult to guess what actually distinguishes good from poor models. Therefore, the atmospheric features are also examined to get some clue. Now, we will see the capability of semify models in reproducing the atmospheric circulation and convergence divergent patterns associated with tropical South Atlantic. Well, this figure shows the regression of JJS seasonal anomalies of zonal and meridional winds at 850 at Tapascal onto the standardized tropical South Atlantic residual. Figure B and C are same as figure A, but for the average regressions of 17 good and 15 poor semifinal models. Well, the observed regression pattern shows that the warm SSG anomalies over tropical South Atlantic is associated with the westerly anomalies along the equator, which is located slightly northward in observations as compared to good and poor models. Well, you can also see that this response is stronger and more closer to the observations in good models as compared to poor models. Well, the observed regression pattern also shows the easterly anomalies over Africa and in the Somali jet vision, which is consistent with the gilt type response due to the heating induced by tropical South Atlantic warming. Well, as compared to good models, poor models fails to show this response and show stronger response over the Indian, over the Eastern Indian Ocean region, which is consistent with increased convergence over India. Well, this weekend Somali jet reduces the rainfall over India through the divergence and due to the reduced lifting over the western parts. Well, the observed regression pattern also shows also shows an interesting feature that is a cyclonic feature in the extra tropical South Atlantic region, which is a part of South Atlantic ocean dipole and this feature is quite well captured by both good and poor models. Well, at upper levels, the equatorial, at upper levels, the equatorial response is reverse, consistent with the baroclinic response that the westerly anomalies over the tropical Atlantic and the easterly anomalies over the Africa and Somali jet region at lower levels is replaced by the easterly anomalies and the westerly anomalies at upper levels. Well, this feature is also well reproduced by both good and poor models. Well, as compared to poor models, the assembly of good models shows strong signal, strong signal over the western Indian Ocean, which may be related with the above normal rainfall over India. Well, the observed wind regression pattern also shows the extra tropical wave frame feature in the northern hemisphere, which is absent in both good and poor models. This extra tropical path may be the cause of the reduction of rainfall over northwest India as seen in the observed precipitation regression maps. As you can see that, that the observed precipitation regression map shows less rainfall over northwest India, which may be related to this extra, this wave frame. Well, this feature is not well captured by both good and poor models and you can see that such type of feature is not well captured over the northwest India. Now, to examine this further, we have regressed the journal wind shear on the tropical South Atlantic residual. Figure B and C are same as figure A, but for the average regression of good and poor model. Both observation and good models shows a substantial and statistically significant wind shear over the Somali jet region, which is consistent with the reduced rainfall over India. In poor models, this response is weaker and maximum amplitude seems to be shifted over the central equatorial Indian Ocean region. Well, on the other hand, if a large scale dynamical western Yang-Mansun index is used, then all CMIFI models and observations shows negative regressions except MPI ESMP. This indicates that large scale responses are not that different in both good and poor models and therefore the regional details, metals are lot for this tropical South Atlantic and ISMART teleconnection. Well, this figure shows the regression of JJ's and number of ADD stream function at 200 hectopascal from ENCEP and CARi analysis onto the standardized tropical South Atlantic. Figure B and C are same as for the average regression of good and poor CMIFI models. Well, ADD stream function at 200 hectopascal is a good indicator of will type response. As you can see that the observed regression pattern shows the guild type quadrupole response, which is up to the sum extent captured by both good and poor models. Compared with poor models, the good model also shows the pronounced and localized quadrupole response over the Saudi Arabia and Peninsula region as compared to poor models. In poor models, it seems to be shifted. The lower levels, the tropical South Atlantic ADD stream function regressions also shows a quadrupole response, which is consistent with the baroclinic guild type response. Well, both observation and good model shows the low level anti-cyclonic stream function over Arabian sea. That will induce the sinking motion over that region and a low level divergence that causes the reduction of rainfall. Well, poor models also shows the quadrupole response, but the center over the Arabian sea extending towards India is missing. And this center is basically present over the Eastern Indian Ocean region and the Western Pacific, which is consistent with low level wind responses. To further verify this, we have regressed the velocity potential regression pattern onto the standardized tropical South Atlantic residuals and figure B and C are same as figure A, but for the regressions of good and poor models. Well, velocity potential is the measure of divergent flow and is often used as a proxy of walker circulation. Well, at upper levels, a region of negative potential has diverging winds that exemplifies strong convection or rising motion at the lower levels. Well, the observed wind regression pattern shows that the warm SSD anomalies over tropical South Atlantic is associated with anomalous convergence over Indian subcontinent, Western Pacific and far Eastern Pacific and divergence over the tropical Atlantic as well as over the Central Eastern Pacific region at upper levels with anomalous divergence and convergence over respective regions at lower levels. Well, the ensemble mean of good models also shows exactly similar feature as being observed in the observations. Whereas the ensemble mean of poor model fails to show the convergence over Indian subcontinent and also the divergence over the Central Eastern Pacific. In fact, it shows the strong convergence over the Central Eastern Pacific and weak convergence over the Western Pacific. Such differences in large scale features may be responsible for the difference in tropical South Atlantic iso-mart teleconnection in good and poor models. To further verify this, we have plotted a scatter plot of tropical South Atlantic precipitation regressions averaged over Indian land points excluding North East region versus the Western and Central Eastern Pacific ocean velocity potential gradients at 200 hectopascal. Well, the negative correlation between the two indicates that larger the Western Central Eastern Pacific ocean velocity potential gradient produces larger negative Indian land rainfall responses. This implies that there is a strong relationship between the quality of reproducing the velocity potential regression pattern and the tropical South Atlantic iso-mart teleconnections in models. So the conclusion is generally all the models show well-defined spatial pattern of tropical South Atlantic, but out of 32 semi-fine models under consideration only 50% are able to capture the tropical South Atlantic iso-mart teleconnection. Well, both good and poor models show large scale responses which is quite consistent in observations. Large scale walker circulation adjustment to the tropical South Atlantic SSDs is identified as one of the factors that accounts for the difference in the low level stream function response. Results revealed strong relationship between the quality of reproducing the velocity potential regression patterns and the tropical South Atlantic iso-mart teleconnections in models in particular with respect to the western and central eastern Pacific ocean velocity potential reagents at 200 hectopascal. Such relatively subtle changes in the response leads to the reversal of the signal over India. That basically raises the question about the robustness of tropical South Atlantic and iso-mart teleconnections. Thank you.