 Welcome to this session on OS climate. OS climate stands for open source climate for those of you that don't already know of us. Today I'm going to cover to you what we're doing in terms of trying to make a breakthrough in open source data and analytics to help solve the climate crisis with regards to accelerating climate aligned finance. So my name is Matt Sandow. I am the chief of staff for OS climate and I'm the physical risk and resilience lead. I've got over 25 years experience in finance in risk management. Most of that working for BMP Paribas Bank and I'm currently seconded by the bank to the open source climate programme. So let's begin by defining the problem. So in simple terms we all know that the world is over exposed to climate risk and we're under exposed to the enablers and solutions that are going to manage this risk and get us through the carbon transition. To frame the problem on very simple terms the way to think about it is that our global growth is around about 90% correlated to greenhouse gas emissions so we have to break that correlation and decouple carbon from the economy. Now in terms of the specific issues we know that there's a significant amount of financing that needs to be driven into the transition. Whether this is for adaptation plans for example refinancing infrastructure, helping oil and gas companies, become green energy companies, helping navigate the supply chain investments that are going to be required to make such dramatic transformation. The other issue of course is that there's also a risk problem that we know as we navigate through the transition and we're seeing growing changes in physical climate events. There is a risk topic and potentially a systemic risk to the global economy. So in OS climate what we've done is we have taken two actions. We took a view that we wanted to build a holistic data and analytics solution and in order to meet the Paris Accord goals and manage the risks. To do that secondly we decided to do that through a co-op and this is leveraging and making a bridge between the research community, data providers, financial institutions and so on there. So that's our OS climate commitment. In terms of the ecosystem, so what does our ecosystem look like? So on the left hand side we've got the community that I just spoke about comprising the data providers, banks, asset managers, research, NGOs and in the middle we have the data problem. So we know we're going to need a significant amount of data to address this climate challenge. Whether that's data on carbon footprints, whether that's data describing climate hazards, whether that's data describing future potential transition pathways. There is a data problem and that data is not going to come from a single source. We believe that data is going to come from multiple sources around the world and that needs to be managed appropriately. On the right hand side the OS climate solution is around three analytical tools and I'm going to drill into these in more detail. Firstly in the top left of that box we have the alignment tools which are tools which help asset managers demonstrate whether the investments they're making are aligned with the Paris agreement. On the top right hand side we have the physical risk and resilience stream which is a programme where we're looking at mechanisms to identify and measure the risk in the system to climate change. So how we address more wildfires, more droughts, chronic heat changes, wind events and so on. In the bottom we are performing also transition analysis and we've released some open source tooling for that which helps the world understand the potential different economic pathways to the right resolution to decarbonising the world. So three core analytical projects and one data one. Our membership, we were founded by BNP Paribas, Alliance, Goldman Sachs and Amazon Web Services and we now have built up a few other asset manager members such as a Mundi, we have a London Stock Exchange, BNY Mellon and then we have alongside Amazon we have a Red Hat who are key partners in particular in the data project. And on the bottom there you'll see also some partners that are research institutions or data providers. So OS Climate we're a member driven non-profit organisation. So if we quickly frame the problem, if you are an investment manager you need to make commitments about decarbonising your portfolio. And these are going to rely on absolutely imperatively accurate carbon footprint data in the corporate domain. So for example if you're lending money to a construction company you need to understand their scope one, two, three emissions. And we also be able need to make forward looking predictions on what those emissions are going to look like. Then asset managers, people managing money are going to need to manage their client reporting in terms of being able to use highly trusted models and tell their investees where their money is going. That also is governed by extremely rigorous compliance rules. In the middle there you've got your risk analyst. So the risk analyst needs to understand the risk in the portfolio. And this introduces the concept which has been introduced into finance in the last few years where we actually have to connect climate data to economic data to be able to make an economic inference on the assets that are being invested in. And on the right hand side we have the procurement officers and regulatory compliance officers. So it's a data problem on the two on the right hand side, data management problem and on the three on the left hand side are very much the use cases around investing. So let's just quickly recap if I'm talking to you about climate change before we dig too deep. I'm not a climate scientist but I'm going to give you a very quick overview of what we talk about when we talk about future projections of climate conditions. The best way of thinking about the way that climate models work is that we break down the earth's surface and then above that earth's surface we create little cubes going up into the stratosphere. And based on various emission cycles, water and carbon cycles, an energy balance is performed and each cube is mathematically solved between one another in order to derive the temperature conditions around the globe and from that models can be used to derive the wind conditions, rain conditions, sea ice melt etc etc and that's pretty much the way that these models are operating. What's fundamentally important here is that these models are looking into the future. So it's not just about a retrospect of what did the weather look like last year. It's how are things going to progressively worsen over time and these need to be modelled under various different greenhouse gas emission cycles. Ok, now you may not be able to view this slide too well but it's not too much of an issue because all I wanted to introduce for you here is the concept that because there could be some financial instability in the system the regulators have got together to define their context of the potential instability that could come as we try to manage our way through the transition. And I think this is great because it really helps everybody understand what we're facing. Fundamentally if you look at that grey line, rising line up in the top left hand side, the thick grey this is a scenario at which emissions continue to grow and the global temperatures rise beyond 3 degrees out beyond the 2050 time horizon. So that's like a do nothing scenario. And then you've got a smooth dark green line that's an orderly transition. So if we get everything right and we model everything perfectly we can enter into an orderly transition which should be the most economically viable and probably the most friendly to the global south in terms of socioeconomic impacts. But what the regulators also looked at and said well if we're delayed in making these transition decisions we need to also take into account the perspective of very volatile announcements in policy and so that lighter green thick line where you can see the sudden downward drop in carbon emissions is exactly that sort of scenario so that's called a disorderly transition. So if you make your way over to the right hand side of that scenario you'll see these four boxes. So in the case of the do nothing scenario in grey we have a lot of physical risk which is the bottom right hand side of this drop graph so we expect to see worsening climate events more wildfires more frequent rain events more frequent wind events and we're not transitioning at all so that's referred to as a hot house world. If everything goes perfectly we end up in the bottom left hand square with an orderly transition with still some degree of physical risk because we know no matter what we do we're still facing rising temperatures at the moment and obviously disorderly will create much more transition risk much more volatility in the system and that's going to be much harder for finance to navigate because the decisions of who to lend to have to be modelled around potential extremely volatile policy decisions and corporate action. That's by the NGFS by the way if anybody wants to look that up. So I said I'd talk through the four work streams of OS climate. Let's step through these one by one. So the sector alignment project is a project that's led by Allianz and if you're lucky enough to have money or have investments or have a pension fund then you'll probably be interested to know where your money's invested so obviously what we're trying to do here is by using carbon footprint simulations we want to try to get a flavour of what's the temperature of a portfolio. So are you invested in a one and a half degree pathway, a two degree pathway or a four or five degree pathway? If your investment fund is full of heavily carbon intensive companies that are not making any effort to transition you're going to be up in that top red orange zone. We all want our investment portfolios of course to be in the bottom zone. Note of course that this is a temporal topic right it's not necessarily about the emissions today it's about capital expenditure, capex commitments in terms of where we expect those footprints to be reducing in the future. So the project, what we decided to do in OS climate is we wanted to open up the toolbox and release open source code that connects to our data mesh architecture and we're able to demonstrate where companies are aligned or not so are they behind their carbon budget or are they ahead of their carbon budget? We believe that we would much rather be releasing this into the open domain so we can have a global collaboration and reach a consensus on the methodology that can be deployed. So a quick summary of this project, this is only a two slider so what we're doing is we're objectively assessing the emission targets of companies and their projections. We're able to compare different decarbonisation scenarios on the platform and we're trying to create a little bit of independence away from commercial data providers that provide this sort of measurement and open up that black box. The benefit for sure is that we think that these computations need to be as transparent as possible. No model is perfect but let's make them as transparent as possible. A complicated slide here but we're not going to go into it in too much detail. The second project that OS Climates is working on is a project that's led by Cap Gemini and this is the transition scenario tool, it's the one I introduced at the beginning. So I've already introduced you the concept of different greenhouse gas pathways, different speeds and paces in which the world is going to decarbonise and we still don't know the optimal pathway that the world is going to follow. We don't know how policy is going to unfold, we don't know about technology innovation, we don't necessarily know the energy mix of the future. So how can banks' investors' lenders make the right decisions of investing and lending in the right decarbonisation zone and how can corporates themselves make the decision of how to transform their entire supply chains without going bust in the process. So the idea of the modelling that's been released, we've released this integrated assessment model into the open domain and the idea is that it's known as a system of systems model and very simply it takes into consideration population, takes into consideration there is a finite amount of natural resources and different energy models are able to be modelled so are we talking about more solar, more wind, more hydro and the macroeconomic conditions around that are then defined based on a different policy scenario. So it's a full simulated engine on a global scale enabling us to make some different scenario inferences of the way we may decarbonise the world in the future. I think I've covered all of that one. And again, I think the concept is that this is, whilst some of it may be commercialised as a commercial service to go and do some advisory work for particular corporates or financial investors firms, the idea of putting it in OS climate is that we have an open layer for input and collaborations from the expert community. Okay, so I've talked about two projects that are very connected to transition risk about the decarbonisation of the world and what I haven't yet talked about is the physical climate risk piece so what do I mean by physical risk? If you think in the most simple terms if you are a lender or investor and you're going to make a decision on lending money to a farm that could be caught up in rising drought conditions, rising flood or you're about to issue a mortgage on a property that's on the coast and could be under water in 20 or 30 years time that's the context of physical risk. And so this is a very serious topic even though some of these issues might only become apparent in 20, 30 years time what we're starting to see is as our ability to model changing climate conditions in the future improves we're going to start to assign, associate that risk into the asset values that we see today and as you've probably seen already in the news there's going to be areas of the world where you can't get a mortgage you can't get insurance on a property and this is going to be certainly an area of growing concern so we have to make sure that our assets are correctly valued around the globe. So if you are a lender or an investor what are your use cases? Well we know we have to disclose our risks and our strategy around how we are managing climate risk ourselves we know we need to be able to measure and assess our portfolio risk under different scenarios because we don't know what the future looks like we also know we need tools on the top right there for origination so how are we going to decide to lend or invest in a particular project finance or the mortgage example I just gave you or the hotel chain on the coast we need tools that will help us make those inferences on investments Operational risks, we also have our own operational risks even in the finance community if sites get shut down because wind takes out the power or wildfires take out the local cellular network the operations shutting down causes a strain on disaster recovery mechanisms and these themselves have to now factor in potential growing climate change conditions also for strategic planning we're not likely to double or triple the size of an office in a location that's likely to suffer more degraded climate conditions in the future and on the bottom right the main ambition of what we want to achieve in OS climate is that if you can measure the risk then you can manage the risk so by measuring the level of physical risk on these assets and potential investments that will lead to adaptation finance so you've already heard about adaptation finance in the carbon transition world so a bank can finance an oil company to invest in green energy or a bank can finance a company to decommission a carbon intensive plant but there's less investment in less money flow moving towards resilience and adaptation so how do you measure physical risk fundamentally we take these building blocks on the left hand side so we start with the climate hazard this is the hazard model that describes in a temporal and spatial domain the probability of an event happening so for example if we take a longitude and latitude in California where there's a power plant we can see the probability of drought or high air temperatures impacting that plant you can measure the exposure by matching those climate models onto the asset locations so you know that my investment is exposed or not exposed to these climate conditions but then there's this really fundamentally important third building block which is the vulnerability piece and the vulnerability piece is describing how impacted those assets will be for example a power plant suffering growing drought conditions where it can't draw water off the river or high air temperatures where it's having to shut down is going to be much more vulnerable than a warehouse or an IT software services company so we try to build a library of these vulnerability models and ultimately if you can put those building blocks together you can build a risk distribution upon which you can make an investment or credit decision and if you think about what I've just explained the majority of those blocks sit in the non-competitive space so we believe that this is rife for pre-competitive collaboration and it's only the final 5% of this modelling that needs to be taken inside a bank's proprietary credit or pricing or decision making systems and this ecosystem sits on top of the data, Commons data mesh that I'm going to explain a little bit later accessed through a data exchange so you can see a couple of examples on the right hand side there where on a simple perspective we can just map out a heat map where an asset may have a larger exposure to coastal flood or potentially riverine flooding and on the bottom right there you will see the chronic heat model that we released a few months ago where we are doing two things with chronic heat we are trying to build a picture for where areas of the world will have to shut down because heat conditions are simply too severe for long periods of time so that's the chronic average rising temperatures but what we're also looking at and you're probably going to hear more about it in the news in terms of the combined humidity measurement so when you actually combine humidity with temperature you come up with what's called a wet bowl temperature measurement and that effectively is the rate at which the body can sweat to cool itself down so if you have a huge amount of manual labour that the workforce is going to go slower and slower and slower and obviously there's a certain wet bowl temperature where you actually reach death but before you reach death your work's going to slow down or probably be disrupted so this is the sort of thing that we're implementing in the open domain in OS climate and here is a wind model that we've literally just introduced this week so this is a great example of collaboration with universities this one is coming from Imperial College in London and we've just taken their wind system model for tropical cyclones which is giving us a feel for where the areas of the world are going to suffer high sustained wind speed risks and obviously as a bank, lender, perspective we can now match those climate hazards onto our mortgage portfolio for example or whatever we need to do if I get a chance I'll demo that but what you're looking at there is an open sandbox it's available to the public so probably a lot of you attending an open source summit will be wondering where's the commercial layer in everything you're talking about so I've talked about the need for financial institutions to collaborate because banks didn't have any climate science knowledge a few years ago and we didn't know how to optimise geospatial files so using collaboration is a perfect case of sharing the workload but what we're expecting to do as well is by building a common coding ecosystem at a base layer we can plug in more advanced commercial models on top of the same platform so what OS Climate is doing is creating a plug and play environment where very precise commercial models can be plugged in just as easily as open models so in the example of what I've just explained to you about a mortgage let's imagine that's alone to a multi-billion hotel chain you can use an open model that's going to give you an indication if that's at risk or not but you're probably going to want to purchase a damage function from an insurer and a more advanced one square meter resolution flood model just to be sure that you really understand those risks before you hit the invest button so it's sort of a modular plug and play environment giving space for commercial data providers to also work with us what we're also going to look at in OS Climate as well is the concept that what I've explained to you today is is all about single asset risks but there's also the concept of catastrophe modelling which some of you may be familiar with which looks at scenario events so if you use catastrophe modelling these are the climate models that describe scenario events for example we can create a massive flood event across Europe for one season and that appears as a scenario block and you can overlay that scenario block onto a mortgage loan portfolio to take a view on whether the portfolio is likely to suffer a big valuation deterioration okay so just a couple of minutes on data commons and data mesh so we've talked about the three tools the data layer has been executed by Red Hat and obviously I'm talking to you from the financial institution perspective I'm not going to be presenting this at a technical level at all but basically Red Hat have taken the decision to deploy a data mesh architecture for this solution and as a recap to the ecosystem of OS Climate you can sort of see the scale of the problem so on the left hand side there you see all these multiple different data sources that need to be federated there's never going to be one central database where everything's stored so we choose the data federation model where you can have a federated governance approach and use the latest technology to then fire those whether they're API-able or other mechanism into the tools that we're talking to you about building so let's drill into a couple of those points so the way that Red Hat will talk to you about this is and some of you are probably aware they refer to treating data as a product that's one of the concepts of a data mesh architecture and the idea is that what your data consumers are considered your customers so if you can frame your tooling and your mesh functionality around individual use cases then you're better meeting the needs for example a data provider on the left hand side here a data provider may actually be a commercial data provider and what they can do through the governance platform is set the rules of who gets to view that commercial data so where does it get distributed to through the federation model and that's fundamentally important in terms of the way we can actually distribute commercial and public data through the platform some of the other principles related to what I've just mentioned is the Red Hat talk about the importance of a self-service architecture and the idea is that this doesn't take a huge amount of coding complex skills in order to replicate models because Red Hat have opened up all the patterns so if Red Hat have a pattern which can be API-able and replicated it's much easier to onboard new data sets in the mesh which is all open sourced by the way so anybody can create their instance of the data mesh the second principle is this decentralised product ownership and what this enables us to do is it's a federated governance model so I'll come to that in a minute actually the decentralisation means that the mesh can keep track of the lineage and changes in data sources and metadata and put that through the same sort of rigorous governance as you would do if it was code so beyond that principle there's also a fundamentally useful principle that because you're federating the data rather than copying it into a lake for example the governance and controls mostly maintained at the original source layer and obviously overlaying on top of that is a federated governance model where you can set some open standards set some principles around how to treat that data in fact we use the data as code context and then that enables you to set a macro level of rolling under which a data provider can set their own rules but within a common framework and Red Hat sort of left me with this slide which I'll probably try not to explain as I'm not the data expert but in principle this is why they like the mesh architecture because really that federated data governance model is simpler to manage and maintain SQL based interfaces and a distributed cloud model which basically is also super useful in this environment because actually there's going to be a lot of data sources that can't be centralized Governments, countries of the world will actually want their data to stay where it belongs so if for example you're the Singapore government you want your Singapore companies to disclose under your particular disclosure rules data being centralized somewhere in Europe or the United States hence the beauty of a federated model you can keep the data where it is and port it through to those people that need to see it and what I mean by that in the sense is that data might be completely public but also data might need to be shared between one supply chain or up and down the supply chain or it might need to be shared between a lender and the investors in my example if you're an investor in Europe you might be given the access rights to see the data that the Singapore company is disclosing and for those of you that are into the open source that's the architecture I'm going to leave you with those slides everything is open source in particular there you've got Trino on the federation side and this is a little summary on the mesh pattern the principles that we're working on is really availability, reliability and comparability and as a closing remark the context of what Red Hat have designed here is data agnostic and so whilst it's actually super useful for managing this extremely complicated climate problem it's just as relevant to the biodiversity in nature a challenge that we have which is an even more immature field and in fact the WWF have picked up on this and have written a very interesting paper describing how it could be a very good idea to deploy the OS climate mesh for biodiversity in nature so that's a summary of our ecosystem transparency is really at our core and very much you can see that we presented to you an end to end solution today that's the approach that we think is much easier to bring in collaborators from academia and having them bridge between financial community and what I would say as well is that I think I'm preaching to the preacher but collaboration clearly is what's required in this urgent call for action 90% of this problem is pre-competitive in our view transparency is at our core biodiversity data mesh could be a very good idea and as a final word we will be in COP later this year but also if you want to get involved we are on the lookout for BAs data engineers, data scientists react experts and there's also opportunities to volunteer which are extending into the public space so for example when I've talked to you about physical climate risk the model that a bank needs when it's lending money to a farm is the same model that can be used in Nigeria to look at the resilience and adaptability of farming of crops over there and in fact we have a project underway today where there's a hackathon looking to empower the youth data scientists in Nigeria using the OS climate ecosystem and adapting our model specifically to the crops over in Africa and we hope that approaches can be scalable as well across the rest of Africa so thanks very much, I think I'm just on time and I'll hand over for a couple of minutes of questions Do you think it's having the correct effect to drive investments into adaptation and transition or could it be a non-dependent consequence of it actually driving investments to be in the safe spaces of it? Yeah, I mean look that's a great question there's a lot to say about that first of all this is not just about risk this is also about opportunities that there are huge opportunities in the carbon transition there's opportunities to make the right loans and to have a very successful business so having the ability to model and understand that is important the context of shying away from risk is certainly an interesting one but the way that a lot of financial institutions would look at it is to say well we want to support our clients in their adaptation so if we identify that a client is at risk because they're going to be caught in the wrong pathway we will be approaching them and saying would you like some transition finance to change your business model so it's also a finance opportunity so you don't really want to abandon your clients necessarily some financial institutions may of course want to abandon those that have made absolutely no commitments at all to decarbonising their portfolio so yeah it's a good question and it's the same question that the insurance industry has about the mutualisation of insurance do you reprice mortgages in areas of high flood risk or do you mutualise it across the entire mortgage portfolio it's yet to be seen but again I would hope that that also leads to adaptation finance so you measure the risk first then you identify the adaptation possibilities so you would hope that as our ability to measure this improves there would be connections to the governments of the world, the regional planners to actually make the adaptation strategies and plans that also need to feed back into the models so the lenders know that there are no longer risks it's a fascinating topic and there's a lot still to be done on it How quickly do you see some of these things happening that already some of the insurance companies are exiting Florida because of the risks there to hurricanes and rising temperatures and they're not waiting for the to see if there's a problem that we're solving today or if there's a problem in 2030? Well I think that through the sort of projects like we're doing in OS climate we can bring some vision and transparency to how the future is going to look which means that we will have the capacity to probably value these assets in the next five years I would say is my gut feeling I think where there will be a reorganisation of risk so a problem in 30 years time is actually a problem for me in the next five years if you're a big investor lender you've got a big portfolio you need to understand its composition and where it could have risks and then the context of how do you support those regions then needs to be the topic of moving to the governments and regional planners and insurance associations looking at it may also be the birth of new reinsurance products because it's going to be maybe a series of catastrophe type bonds which actually might allow the governments to still provide the insurance buffers in some of those areas so I know for example in the UK the insurance buffer orchestrated by the government runs out in 2039 and at that point there could be 300,000 properties that are no longer insurable so something's got to happen and it's the same thing we're seeing in America as well again another really interesting topic and hopefully we're bringing it to the forefront there's already existing efforts like IPCA which one IP not sure about that one but we are in terms of the climate model in climate science we're mostly onboarding pre-reviewed models we're trying not to build our own the only thing we're doing from time to time is we are restructuring or packaging files that are downscaled from NASA for example if you're into your climate topics when the IPCC released their latest series of climate models known as AR6 and that's a brilliant report if anyone hasn't read it they NASA took hold of those data sets and downscaled for example the heat and drought indexes and we've taken those and structured them in the way that can be applied to finance because the problem was today is even though it was just a few it was a few years ago when actually economists and climate scientists started working together so if you go back 15-20 years everyone was working in silo so we've already seen one wonderful thing and now we're sort of by having an open collaboration like this that bridges world class technology financial institutions research we can actually start to collaborate much faster with research institutions because they can see those practical use cases because like I said a lot of the models are just not usable today to make those decisions so we have to restructure and play with them but what we hope is that the world sees that and then the research community adapts the way that they are thinking about models in particular around the concept of uncertainty because that's not necessarily something that's thought about in the initial provider space today where you'll get an answer of X if you purchase a climate model from somebody whereas what we want to see is the uncertainty behind those measurements and be able to get a flavour for what that means so yeah it's a great area to be in but I think that we will probably steer clear of doing too much modelling directly by ourselves we'll stay at that sort of package level where over time so I'll stick around for a few minutes if anyone wants to grab me afterwards but thank you very much everyone