 for this very inspiring talk. I'm Benjamin Sulton. I'm a researcher in France, in the south of France, in Montpellier. And I will mainly give some example of what Alex just shown about climate change impact on crop productivity, on crop yield, but mainly with a focus in West Africa and on rain-fed crops. It's quite limited. But it could be very important for food security, actually. So Africa is particularly exposed to climate variability and climate change as well. You have one-third of the population facing widespread hunger and chronic malnutrition. And you have also people whose livelihood is heavily dependent on traditional rain-fed agriculture. So there is a high vulnerability. And climate change could pose an additional burden in achieving food security goals in the region. So indeed, when you have additional increase in global warming, when you have changes in hot and cold temperature extremes and also changes in precipitation, you could have a strong impact on crop yield. And you have here projected changes in Africa with an increase of four degrees of global warming. And you have on the left annual maximum temperature, annual minimum temperature. And on the right annual total precipitation and maximum daily precipitation. And you can see that there are strong warming in Africa, especially in continental Africa. And it's very true in the Sahelam. And you have also changes in the rainfall. But these changes are not that uniform. If you look at annual precipitation, you have part of Africa, especially the south of Africa, but also the western part of West Africa, which is drying with the increase of temperature. And some other parts who are on the opposite with a lot of more rainfall in the future. But you have a common feature in Africa, which is the increase of the extreme rainfall with a higher number of maximum daily precipitation, which could also be a problem for crop yield. So this change in climate, especially the warming, as has been shown by Alex previously, could lead to negative impact on crops, especially because of this higher temperature, which could shorten crop cycle lengths with this index of growing degree days. And if you have some more temperature, you have less time for crops to grow and to produce biomass and yield them. And you could also increase water stress with this higher temperature, which is overall detrimental for crop production. And overall, we could expect less yield, but also more viable yield under warmer climate. So as an example, this study shows the distribution of crop yields under present and future climates based on semi-fiber stimulation. And it is crops from the crops we use here, are millet and sorghum, which are the main staple food crops in the region in West Africa. And these are anomalies given from a crop model. So you can see that the distribution is shifted in future climate under lower yield, with also a higher distribution, which means that you have more viable yields under climate change. So such kind of projection, we designed a portal with the National Med Service in Senegal to disseminate such kind of projection of climate, but also projection of crop yield that could be used for decision making or at least for sensibilization about climate change impact on the crop yield. So you have here the portal we designed with colleagues in Senegal. And you have, for instance, here different kind of indices we put on the portal with temperature, with precipitation, but also a projection of crop yield. And you have here evolution of maize under different scenarios of climate change using semi-five scenarios. And now we are working on extending the portal with semi-six data, but also to develop the portal to other countries. Here you have an example of what we are doing in Burkina Faso as well. So climate change is not climate change and climate warming is not only in climate models and in future scenarios, but it's also visible in climate observations. You have here an example of what we have in the last IPCC report in the Regional Fact sheets in Africa. And you have already in the observation mean temperature and hot extreme that have emerged above natural variability. And you have also several rapid changes, some faster changes that we have on the global average with also observed increase in hot extremes and rainfall extreme as well. In this slide, you have the observed temperature in the Sahelm since 1950, during the hottest months of June and in annual mean on the same graph. And you can see that there is a clear increase of temperature, especially during the hottest months in April, where there is a plus 1.4 degrees warming since the 1950. So it's quite important, especially when you look at the hottest months during just before the months and season. So now the question is how such change in terms of temperature has already affected agriculture. And this is not an easy question for at least two reasons. First, when you look at productivity time series, there is a lot of variability and trends in the productivity time series that are not due to change in climate, but that could be due to management change, that could be due to a lot of factors that could influence clock yield by the end. And second, there is also high natural variability in a historical climate, especially in Africa. And sometimes it's very hard to attribute the changes in terms of anthropic global warming. And as an example, you have here an example of such variability in Africa with two recent climate extremes in Africa, which had a strong impact on food security. And these two events were reported by the World Weather Attribution Initiative in Madagascar. And for the first event, the Madagascar was facing a severe food crisis exacerbated by exceptionally low levels of rainfall over 2020 and 2021. But the attribution analysis based on past data and climate summation concluded that factors other than climate change were the main drivers of this food insecurity. And in fact, this higher, the high drought belonged to the natural variability of the climate in the region. Another example is the increased rainfall associated with tropical cyclones. And on the opposite, the attribution study concluded that there is a close link with climate change in the region. So you have a high viable climate and it's very hard to look at the cause and to attribute it to climate change. So to answer this question and to assess the historical impact of human activities in West Africa, we designed modeling experiments based on two components. The first component is based on the climate model to simulate historical climate with and without a tropical influence. And a second component based on crop modeling to assess if entropy warming has already affected crop yields in West Africa. So considering climate simulation, we used a global model, an atmospheric general circulation model forced by a SST, by sea surface temperature. And the model was used for two kinds of simulation. First, a factual simulation with actual condition that are influenced both by human activities and natural forcing. And then encounter factual climate simulation with the non-warming climate with pre-industrial climate that lacks, in fact, a human influence on a global system. So here the sea surface temperature and sea ice were retrended so that we could remove the influence of CO2 of human influence in the simulation. And by the end, we got 60 years of IGCM simulation and with an ensemble member of 100 simulation. And so this data were interpolated at 0.5 resolution and also bias corrected so that we could use the different variables to force crop models. And this kind of simulation were already used to assess changes in terms of temperature in different regions of the world. In a paper made by Shogama et al. And also Izumi et al. in 2018. And you have here a result with historical and observation in black. And you can see that the simulation of the model are quite close when you use the simulation using entropic emission of greenhouse gases. And you can also see that the simulation have been used for assessing climate change impacts in the global scale. So using the Sigma crop model so this study based by Izumi et al. I investigated the change in terms of average yield associated with climate change from previous levels. And in red, you have areas where climate change decreased yield on the historical in the past. And on the opposite, you have in green the areas where climate change increased yield. And you can see clearly that the areas where the productions were higher because of climate change, it's very clear in Northern latitudes in Europe and in Russia also in Canada. But you can see in the tropics in West Africa and India, in Southeast Asia, there is a clear decrease of crop yields because of human activities. So what are the impacts of human activities on West African climate? So we have a look in the simulation and we compute several user-relevant indices that could be very important for crop productivity by the end. And these indices are part of a longer list we established during meeting with stakeholders in Senegal and in Burkina Faso. And we then compare these indices so that our annual mean temperature if you rainfall events and rainfall intensity we compare these indices between the two simulations with and without greenhouse gases. And what we found is that most of these indices were significantly different from one simulation to another, especially annual mean temperature. And you have here an example where you show, I show you a time series of annual surface temperature in average over West Africa in the observation. So it's the observation from the data from CRU data. And you can see that there is a clear increase of temperature with an average temperature of 27 degrees over the last 10 years of the simulation. And you have now in blue the counterfactual simulation and it is clear from this simulation that we could simulate the internal variability of the temperature, but we missed the increase of temperature and also the mean temperature is completely different from what we observed in the previous years. And now it's the factual simulation when we introduce now the effect of the anthropic greenhouse gases in the climate model. And you can see that now we are very close to the observed temperature and we also have a very close average of temperature over the last 10 years of the simulation. So now we have done the same for the same comparison, but for mean temperature, total rainfall, very hot days and very heavy rainfall. And what you can see is that it is very clear that we have differences in terms of mean temperature, but we have no differences in terms of total rainfall, which means that we cannot see any signal of increase of greenhouse gases in the changes in terms of total rainfall, but we have some changes in terms of very heavy rains in the model. So we have significant changes in terms of very heavy rains which could also have an impact on crop year. So now what are the impacts in terms of crop years? So we use the atmospheric GCM outputs to force two crop models that are Sarah and Sigma. So Sarah was developed in Montpellier and Sigma was developed in Tokyo and they were used to simulate crop years using these two simulations, the factual and the counter factor simulation. And we have done 100 simulations for each. So let me briefly introduce you the two crop models we use. So we use the Sarah model, which has been developed by Sierra, and it combines a water balance model, so simulating water demand, salt water availability, but also a carbon assimilation and partitioning model with radiation use efficiency with phenology. And this model was used several times. And it seems that when you compare it to a FAO yield, it captures quite well the variability of crop yield in the region. And we also used the Sigma model which was developed by Naro in Japan, including much more processes that what we have in the Sarah model. So you have this growing degree days simulation, but you have also a different kind of effect, especially the CO2 effect on crop yield, but also a lot of different kind of processes that we don't have, such as the nitrogen deficit, heat waves, and excess water that have been considered in the model. So we have first validated the two crop models compared to the observation. You have here the two simulation of millet and sorghum for the two crop models in blue, the Sarah model in red, the Sigma model. And you can see that you have one model doing a good job in terms of simulated the annual trends. So it's the Sigma model. And another model is the Sarah model, which is doing a better job in simulated internal variability of crop yield. So now these maps are the geographical impact of crop yield associated with historical climate change relative to a non-warming counterfactual condition. So when you have negative values, it indicates that you have a yield loss due to entropic warming. And what you can see is that you have a high crop yield losses, especially in the north of West Africa in the Sahel, which is the same in the two crop model. So even if we have two very different crop models, you have kind of exactly the same geographical patterns with crop yield losses due to increase of greenhouse gases and to a historical climate change. And these losses are estimated between 10 and 80% for millet and between 5 and 50% of sorghum, depending on the crop model. And it could be very important for some countries. For instance, if you look at Niger, or if you look at Senegal, we have very high losses because of the increase of greenhouse gases and its impact on climate change. So to conclude, I'll show you some example of the impact of human activity on regional climate. So in Africa, we have warmer climate with more intense rainfall, which could have a negative consequences on agricultural production. And it's already visible when we perform attribution study. So without entropic warming, crop yield could have been higher in West Africa so between 6% and 50% for sorghum and between 11% and 80% for millet, which is quite high and could have an impact on food security. And even if we use two different crop models with two levels of complexity, we have kind of a very similar effect of climate change on crop yield, which suggests that we have a common mechanism which might explain this common behavior. And we suspect that it's likely the increase of evapotranspiration-led water deficits and also the shortened crop duration induced by the warming. And since the most optimistic climate change scenario do not lead to warming below 1.5 degrees in Africa, so we expect further crop production losses in West Africa and it's clear that we need to think about the most effective adaptation methods in the region that will be really critical. So I think I'm done and I'm open for questions if you have. Thank you very much. Let's give Benjamin. Thank you. Benjamin, that was great. And we had a discussion just yesterday about the ways that we might assess things like losses and damages and, you know, the impacts attribution, detection attribution world has become much stronger on the climate side in recent years. But getting to the impact side, I think your study with Toshi is one of the earlier ones that's really doing that, so very exciting to see. Any questions in the room for what we've seen? Yes, let me get the microphone. And just for Benjamin's sake, could you please just say your name and institution? Hi, Ben. I'm Trino from Australia. I have a couple of questions about your presentation. Yeah, the first question is, you mentioned the 100 Inzember members. So I just wonder when the Inzember member is from the different climate model or different initial conditions, so I'm not clear there. The second question is, your work is under which emission scenario and the last small question is, how much confidence you can put on your simulated results? Thank you for the question. In fact, the 100 simulations are perturbation of initial conditions. And so we try to look at the effect, the internal variability of the model and it could be quite high, so we need to resimulate the model 100 times. But we believe that, in fact, if we want to really sample the efficiency, the uncertainty, maybe something like 20 to 30 simulations would be useful by the end. And so the question about the uncertainty of this simulation, I mean that it is related to the portal we have done or to the number I give by the end. I don't capture very well your question. It's about the uncertainty of the whole study or the number I give you by the end. If I paraphrase it, in terms of, how do you measure uncertainty? Do you measure it across those 100 simulations? Do you measure it with other uncertainty elements like the error bar around representation of different processes? So when we saw those box and whiskers, what was the variation in there? Must have also been space? Let's see, I don't know if Benjamin is frozen. Okay, try again. I was offline thinking. Can you say it again? The question is how do you measure uncertainty? Is it across the simulations? Do you have any kind of larger error metrics about how well the model is performing or factors that might be left out, for example? Yes, what we are doing is we are trying first to validate the model across against observation, against yield data, and also we are validating the way the model represents the relationship between climate and crop yield in the observation. For instance, we are looking at correlation between temperature and observed yield, rainfall and observed yield, and we are trying to look at how crop models are doing that and we are confident in the model that they could do that. But it's true that there is a high uncertainty and I think the AgNIP ensemble simulation, you have shown just before a very good example of how we could measure this uncertainty. And I think using different crop models it's a way to sample uncertainty. It's really important to do that. Yes, so just to add on to that, within AgNIP we have taken this study approach that Benjamin and his colleague Toshi Izumi from Japan have pioneered and we are now bringing it to the larger AgNIP community. This is kind of what we do, which is when somebody has a nice study, we say wouldn't it be great if all of the modelers did this because now we could really understand the model uncertainty component of that. So we are actually at our AgNIP global workshop later this month. We are bringing many modelers together to redo this protocol with more models and more sites in Africa so that we can start to factor that in. Maybe one last question in the room before. Can we leave Edmund sometime as well? All right, I don't see a burning hand. So Benjamin, thank you so much. I hope you get to listen also into Edmund's talk. So our next speaker is Edmund Toten, who is joining us virtually.