 seconds. Yeah, sure. Hey, you're live. Okay. Hello, we've already to this Climate Action Accounting Meeting. It's our Genome Meeting happening every other week. Today we have a special guest in our meeting, Deval from Mojo Global. He's the Director of the Technical Student Committee there. Mojo Global is a collaborative project under the Linux Foundation and they support ambitious climate action by bringing together a community of experts to develop open source software. Deval is presenting their flagship project called Flint. It's full lens integration tool and it's a solution for MRB, so it's monitoring reporting verification. This is a hyper lecture call, so the antitrust policy note is in effect as well as the hyper lecture code of conduct, so please respect each other, be kind and friendly. And at the beginning, we always want to give new members a chance to introduce themselves. Is there anyone in the call who wants to say hi and provide a little bit background information about themselves while they join this call? I'll jump in. Hi everyone. My name is Tim Lip. I'm from Calgary, Alberta and I'm the CTO for a social enterprise that represents a bunch of different local farmers and distributors. We're really eager to better track and report on our carbon emissions and the reductions we generate with our practices. So really excited to hear more about the conversation. Great. Thanks for joining Tim. Anyone else? Okay, Deval. So if you want to, you can start sharing and kick off your presentation if you're ready. I stop share. You guys see the screen? Yes. Yes, we can see it here. Perfect. All right. Thanks. Thanks everyone for joining this call and thanks for inviting me. And as Robin mentioned, we'll spend next few minutes talking about the flagship project from Mojo Global. It's called Flint. It stands for full line integration tool. And maybe I'll start with a bit of a background and then dive as deep as we can without, if you're anticipating code, pulling up GitHub and code, you'll be disappointed a little bit. But I think it's important to discuss why and how to design this MRV system and what was the motivation behind Flint in general and how it is probably different than what are the things out there. So a little bit about myself. I know this is a Hyperledger SIG and I have to start by a disclaimer. I know nothing about DLT and probably the better way is I know enough to be super dangerous. And so I claim to know nothing about DLT. I do know a little bit about land use, land use change and how it affects into carbon accounting. But I'm not a soil expert or I'm not agronomist. So not an expert, but I know a little bit. I know a lot about artificial intelligence and machine learning and its application in climate action. And fortunately or unfortunately, we are not going to talk about it in this call. But if you want to talk about that, I'm happy to have a conversation later. So let's dive into this. So what are the expected outcomes for this session is I have to say I shamefully stole a bunch of slides from much longer three day course which Mojo Global was delivering at UNF CCCC earlier last month. And so we can skip on. So anyone wants to raise their hands and say, this is not useful. Let's go to the next section. I'm happy to do that. I'm also happy to take questions. Stop me in the middle and we can discuss more things. The idea is to have probably a discussion rather than going through the slides. So what I would want to cover is some of the impacts of policy drivers on designing a technical system. And how do you design, build and run a system for reporting and projections? There was a whole section around how do you engage with open source tools, but I'm pretty sure this community doesn't need to know that. And so we'll skip that one. And then there are options of what is Flint, how are, maybe give you, I do have a couple of videos which will show some of the examples on how people are using Flint. And then what are different datas and models and options available in designing the system? So firstly, like Flint and Mojo has been in the making for a while. And so there are like a bunch of different contributors. It's an open source project. We have currently we have 260 active contributors on the project. We are cousins in a sense because we are also part of the next foundation and let's sit together in that foundation. And it has been a work in progress. So Mojo was officially launched six years back, but even before that, what makes Flint right now has been a work in progress for decades. And some of our key advisors and authors of the Flint system are also lead authors on the policy recommendations and for UNFCCC and IPCC reporting and all of that stuff. And then there are a bunch of different, we have a bunch of support from different governments, including a government of Canada. I heard Tim is in Calgary. One of our chief scientists is in Calgary and so happy to introduce you to him. And especially like even Linux foundations and so on. So what are different? I think the first place to start discussing anything around like MRV is around policy drivers. And this is what we feel. So I think policies are very regional. They vary from country to country. They also vary from region to region. And not just what policy means for the system, but what does this output from these systems can help in designing policy. So it's a two-way street. And so to understand what are the policy drivers, what do we want to achieve is probably the first place to start when we start designing an MRV system. And I'll briefly touch over just two concepts. And a lot of you might be familiar with them or probably not. And so the first concept, and this is nothing new, it comes from probably from 2006 IPCC guidelines and it says this concept around say tech, which is like transparency. And transparency doesn't mean it has to be simple. It means that it has to cover everything. And it has to be very open about what it can and cannot do. And some of that could be very, very complex. So one is transparency. The other one is accuracy. And this is a tough one because how you define accuracy will depend on your methods and we'll quickly talk about that in the next slides. But it has to be accurate. It has to be consistent. It has to include, it has to be as complete as possible. So it has to include all different pools. It has to include different ecosystem parameters and policy parameters and so on. And it has to be comparable. This comparability was, it was not originally in the red standards, but it's now in the NDC guidelines, which is slightly newer concept, but it has to be. So that's why the ad is seeing it. There are two major things based on guidelines, especially IPCC guidelines and around MRV system, and which is good to understand because I'll be referring to those going forward. One is peers and the other one is approaches. And in tier one, for example, is a default. It's at like the global levels. These are emission factors, which are, there's obviously science behind it, but it's a very, very macro level and you apply those. Tier two is more country specific, right? And country specific, geography specific, comparatively more accurate. Tier three is more advanced, even more accurate based on models, based on ground measurements. It's more flexible. Even in some cases, it can be based on the land parcel by land parcel that's looking at more, more, more information there, right? Same thing with approaches and not to confuse tier with approaches, right? So like approach one is what's the total land area for that period. So there's no change in formation. It's a point in time and says, okay. And most of it is estimates, right? And which is fine. Approach two is around net change over two periods. And which is to say that what happened between let's say 2020 to 2022 or 2010 to 2020. And it's between those two periods and what's the net change. But it does not talk about what happened in between those. And so that's the second approach. Approach three is most especially explicit. It talks about cross changes. It takes into account temporality explicit measures. And so say there was a forest, then it was used for agriculture, but then it was converted back into the forest. All of the temporal explicit data is considered in that as well. Now the two things, the steers and approaches both are actually not based on inputs. They are based on outputs, right? So just because you have national forest inventories does not mean that you're using tier three. It depends on what outputs you are using that to create. Same thing, like say if you're using maps, it doesn't necessarily mean that you're using approach two or three. It could be different, right? And so tiers and approaches, this is I think a more disclaimer, but tiers and approaches are not mutually exclusive. And a country or a project can use a mix of these different regions within the country, within the project, right? So you could have a tier one approach two or tier two approach three and all number of combinations of those when you are reporting. As long as again going back to tech, which is as long as you're transparent about like how you are doing that, right? So that's the part of transparency. The big guiding principle need for Flint was can we run this different approaches and different combinations of approaches and tiers using different spatial and a special activity and management, all different types of data. And that is why Flint was built, right? So that's why there's a huge emphasis on that integration piece and we'll keep on going back to that integration piece as the backbone of what Flint can do with this data. So yeah, I think this is, I like this table again, shamelessly stolen from IPCC 2019 guidelines. It talks about different methods and approaches, right? And I don't want to go into like each and every kind of box in this table, but so there are wall-to-wall methods, there are survey-based methods, and you can track when you are looking at say across three people can with wall-to-wall methods, you are tracking pixel by pixel land unit by land unit using consistent time series data. And so it can become from all the way very simple to highly complex systems, depending on the needs, depending on the policy drivers and depending on the outcomes that you want, right? Again, this is probably from a previous report, but it just talks about like there are basically seven recognized carbon pools, right? And so there's above-ground biomass, there's below-ground biomass, and so the more pool, and this is just say top level, right? So you can go even deeper into it, right? And say within that pool, biomass, what are different types of that biomass? Within soil-organic matter, how do you characterize different soil-organic matter? And it can be, you can even go deeper into those pools, and a successful MRV system has to integrate, has to have this ability of getting transparency and visibility and information of every single carbon pool that's available based on the use case, right? So to keep in mind, basically IPCC guidelines are like foundations, right? So it's not, they don't tell you how to build this system, they tell you what those systems should be built on top of, right? So if you have a strong foundation, you can build different types of systems and so on, but they are foundations, they are not guidelines for building the guidelines for the foundations. And obviously, if your foundation is strong, it could mean that you will build a strong building on top of it. Tears and approaches are results of decisions, and they are not really decisions themselves, right? And so in some sense, they are what you select, so you could argue they are decisions, but they should be seen as results of decisions versus decisions themselves. How you use your data and how you model things in your system determines tiers and approaches and not, and this is again going back to the comment that it's dependent on the outputs and not inputs, right? So how tiers and approaches are defined as inputs, right? You have to, I mean, we went through carbon pools and different, I want to emphasize an example here, right? So this is what I mean by like it depends on the output. So both of this picture here, they are actually, they follow tier three, right? But the resolution is different, right? And so depending on your policy drivers and use cases, you can have different resolutions. Again, so if you have a very highly high resolution data, doesn't always mean that tier three. And in many cases, if you have very high resolution data, it also means very high computational costs. And so you don't need that, right? You can get to similar results or useful results, even without having that kind of resolution. So how, any, should, I want to take a pause, any questions that are popping up, anything, or should I keep going? I will, I have a question. Yeah. So with this system, have you used it at all for carbon credit analysis and tokenizing carbon credits or trying to get further financial support through what you're doing with this analysis? Yeah. So for carbon credit analysis, there are use cases where, where it has been used for carbon. And I'll show, there's a very interesting whole that thought there's a use case, but it doesn't. So this is nonprofit. This is how people use it. So people have used it for carbon credit analysis. To my knowledge, they haven't used it for tokenization. And, but it is mostly for analysis, right? And say, okay, how much carbon credit could be within the given policy realms, how much carbon credit or what is the potential for generating carbon credits within a given then parcel. And those kind of analysis have been done using the system. There's actually a really good example in the video, which will follow. So you probably like that a lot around that. Cool. Thank you. All right. So let's keep going around systems design and policy. And so like, the systems design are evolving, right? So that, that's, that's, that's, that is why you have to start with policy, right? So now there are, obviously, there's like NDCs. Now there's enhanced transparency framework. And which is something, I guess, a lot of us are following right now what comes out of like COP26. There is, I think if I'm not mistaken, article six, which is around trading based on and verification and based on enhanced transparency framework. And then there are also some new, like the New York declaration around like tracking restorative activities, right, against the targets and stuff like that. So this, there's for every national system, there are new kind of regimes, which are being brought up, there are new agreements, declarations, and so on, right? And then there's also private sector reporting, which a lot of other people are interested, which is like voluntary reporting kind of schemes. And so you have to start when you are designing the system. This is just to highlight that, you know, understanding the policy is, and reporting needs is, is big, right? And that's what drives this, this system, right, per se. And so I don't want to go into like every single bullet points here, but what I want to highlight is there are policy and reporting needs on one hand, but then there are operational needs, right? So say, how often are you going to run this analysis, right? What is the version control for it? What are the guidelines and guardrails for quality assurances, right? Does the system need to be available 24 seven, right? And these are more operational needs. And this combining this policy and reporting needs with operational needs is what will guide how you design the MRP system. And this is probably quite simplistic, but at the end of it, what you want is you have different types of data and say forest data and other land use data, remote sensing data. You bring all of that data and you're creating products, which is to say you are doing some analysis, building some models, which are giving you some results. Now, all of that draws us how to be integrated. And then relevant information, which is where it is produced, right? And this is how a system operates. And if I put for a second, I swear this is the only machine learning reference I'll do here. This is like the forward pass in your machine learning, which is once you have a model, you there's one way information that say you pass the information data into the model, it gives you an output. But when you train that model, when you want to build a system like this, you have to start from from backwards. So you start in machine learning, you start with say, if it's a supervised learning, you start with like the labels and what is the output that I want, right? So even in this MRV system, a good way of thinking about it is what is the information that is needed. And how can I enable this integration between all the different products which are produced from different types of data, which will then guide my data providers or my product providers on what needs to be done, what needs to be supplied to have a successful integration. And again, this is an adjustment example where we go back to integration. I think integration is the key as you see here, like, how do you integrate all of this together is a key in building this system. So what makes a system operational? Again, going back to and this is the definition which I like and happy to be challenged on this definition of what operational means, but it has to provide necessary information to those who need it when they need it in a standard that is needed. So this is very simple. So in this case, who could be land owners, could be private sector, could be governments, when depends on the analysis, are you doing, say, every year, do you need it ad hoc on demand? Do you do it every month? And what are the different standards here? The biggest challenge here, there's a lot of good science and that has been progressed. If you are in many parts of the world, there's a lot of data, data qualities and issue, but the data availability is not. But how do you move from science to operations is the big challenge. How do you move from taking, say, this equation, this model and operationalize them is the challenge which, and that's what was a goal for Flint, how we can operationalize all of this different publication based science and everything like together. Again, this sounds very much, say, understood like I think people recognize this, but it's good to emphasize that. So the module global and Flint is made up of, it's all about people. So there is technical part of it, but then we have administrative office, we have program managers, we have people who are helping us with documentation. And we have obviously have science experts. And so it's not just a technical challenge, it's way beyond technical challenge where you want to operationalize the systems and actually use it. So developing, now getting into a little bit of how, and this is probably our view on how do you develop and operationalize an MRV system and again, happy to be challenged on this, but what we see is that three major pieces. So one is policy. So the policy sets the need and timing and responsibility and so on. The second is a very strong governance. This is what manages your processes, your systems, defines your rules, provides adequate system resources. And the third one is technical, which is actually like how do you bring this together and produce the outputs that are required, run the models, and so on. And so we will represent this three, like policy with green governance with purple and technical with red. And for those, if to be inclusive, like the policy is the first one here, if there's someone who is colorblind and cannot recognize colors, policy is the top one, then the middle one here is governance and then technical. And usually what people, when you start, you are starting with a lot of the mistakes which we have seen. I think Sherwood has his hand up. Sherwood? Yes, I was really interested to see just the focus on governance. It's an area that we're definitely interested in. And so I was wondering if there are resources that you could point us to so that we could see how your approach to governance in some of this work. Yeah, absolutely. So I'll send some links, Sherwood. Also, I'll briefly talk about governance in the next couple of slides. But in more details, I'll definitely share more resources on that. Thank you. I wouldn't call it mistakes, but for the need of, say, speed, most of the time what happens is people think of the systems in a linear fashion. So say, this is the policy requirements, we send this governance, my end goal is producing a report which will be then submitted to X, Y and Z. And they think of it as in linear terms. And most of them what happens there is it is prone to errors. A lot of it, I have seen a lot of it is done on Excel sheets and difficult to track. And then it's one-off and difficult to reuse. So what we need to do is think of it as a continuous process. And in either ways, and this is how MRV system, in my opinion, should be approached is a continuous process which keeps on improving as we can move forward. So it has to have this ability to evolve. And that's one of the things which was foundational when we were designing Flint. So you have to look beyond near-term outputs, but focus on longer-term process and requirements and how this system can evolve with evolving processes and evolving requirements, which we are experiencing right now. And an example could be, say, let's say now you want to design a process which has some key milestones which could be annual and depending on when you want to do that. And so then system needs to do certain, say, hit certain milestones at a certain point. So you have to complete your annual plan. Then there will be new data coming in. And then when you have new data, then you have to run the system again. You submit your inventory, there's assessment, and then again you are back at, let's say, next year's plan. So I can start with that. In this, the way we see, and again, Sherwood going back to the governance, why governance is this foundational and very important to us is, as you see in this slide, what I want to highlight is the action points and the involvement of policy and technical is not around throughout that cycle. And there are certain tasks and certain deliverables which needs to be there, but governance is present throughout the cycle. Governance is what holds this whole thing together. And so governance is fundamental to this process. And so a good governance includes things like clear objectives, roles and responsibilities, pathways of engaging, decision-making frameworks, and so on. And there could be, there are various, this is just like, say, I highlighted some key milestones, but then there are various things that happen. Let's say when you have a new data, you run that data, you build new models, now you have to integrate those into the system. You do rigorous testing and review of your data and model, then there's Q and QC process. The system has now evolved, so you have to kind of finalize the system design for that time frame, then you run the system and so on. So I don't want to go into each and every one of them, but there are multiple things which actually go into multiple tasks and deliverables, which happen throughout the process. And so there will be times where you will be rerunning this systems multiple times within that key kind of decision-making or deliverable points. And so then operational MIR system should be able to complete this, big one is complete a QAC, and you might be running this models again and again. It has to run the systems, it should be able to assess uncertainty and as I said, it should be able to evolve and also continuous improvement. It should also be able to, let's say there was a change in policy and if you input that or now you have better data and you're going from say tier one to tier two, all of that it should be able to quickly assess those changes and it should be able to run. So it should be able to do what it is supposed to do, that's probably a given I guess. And so the role of Mojo Global here is to provide the software, which is Flint, which you are talking about right now, and tools and so there are various tools around that software and the know-how of how we can build this MRB systems and we do that through open source and then there are various ways of engaging and then these tools are reusable, they're configurable, they meet national. So the system was designed primarily for national level accounting and so it meets national MRB needs but what we have found out is even on project by project basis, it will meet most of the needs there. Robin? Yes, I have a question that was on the previous slide, it was about publishing the data on a public database. Is there one database that fits all or how do companies publish the data and is it accessible for everyone? Hold that thought. So there is a database which is accessible to everyone and it's not by any means complete. We do, if anyone wants to contribute their data, we obviously highly encourage that but even when you get started and that thanks to UNF CCC, we do have tier one kind of database and rules and everything and so we are able to make that available and that's evolving. So that's continuous to get adding to that, depending on how soon we get new data. Yeah, perfect. Can you trip the links to the database? Or just send up the words? Thank you very much. Yes, you might already be speaking to this but I'm curious about the governance around the difference between measuring and verifying. Do you use the same tools to measure and verify? Does it have to be a third party? What's your approach to that for the verification? Yeah, so ideally in ideal case, so tools are very much similar. So what you measure and what you verify, the tools could be similar. They don't have to be. Ideal case scenario is you want different bodies doing measurement and verification. So let's say, even if you are using same tools to get that data, different people should be able to verify the outputs. In many cases, I have seen that's probably not the case. So I think the answer is tools are the same in many cases. But the focus is, let's say you want to be just focusing on verification, then you can use plain and you could bring in different data sources. And you can have different models to verify the results as long as you are following the guidelines, right? As long as you are using, and that's where transparency is very important. Let's say, if you are using these factors, if you are using this geographic area, if you're using this time series, then you should be able to verify that, right? So like, but I think good question. Tools could be same. Systems could be slightly different. Hopefully, the entities are completely different. But that's not always the case. Thank you. Yeah. So video time now. And with all of this background, just want to give a brief introduction to Mojo Global. I couldn't do it better than just showing this video, which is a small two minute video. Let me know if you can hear it, right? So, yeah. In today's world, it's almost impossible to imagine we lived in one without GPS without technology to guide us in the right direction and help us get where we need to be in the world of nature based climate solutions. The full lands integration tool, otherwise known as Flint, is that guiding technology that can show you how far you've come and how much further you can go to achieve your climate goals. Let's start with how Flint can help you know where you are and keep you headed in the right direction. Flint is a modular, flexible open source software that adds a new level of sophistication for an incredibly diverse group of users from governments to private land managers and other businesses and stakeholders. It enables you to measure, report, verify, even forecast greenhouse gas emissions and removals from forestry, agriculture and other land uses. The government agency responsible for annually reporting national greenhouse gas inventory in Indonesia is going to have different needs than a private forest owner in Chile. That's why Flint provides a framework for all its users to develop their systems, data products and capacity in a way that works for them, starting with a simple system that uses the most basic data and emission factors in the public domain, then continuing to add improvements as capacity expands and better data is available. We can't reach the goals set at the Paris Agreement without nature-based solutions, halting deforestation globally, restoring vast areas of degraded lands all by 2030. That's where we want to get to. Just like GPS has come a long way to not just show you the roads but help you avoid traffic and speed bumps along the way, Flint is the ultimate open source software that can help you reach your climate goals in the forest and land sector. Working together, we can help you get there. Go to moja.global to find out more and to see what more you can do. Yeah, so I mean I wouldn't spend too much time on why open source. I think Sherwood, I think you have your hand raised and I'm not sure if this is a new question or it was. Sherwood, if you don't have your hand raised, I can't figure out how to raise my hand and I want to ask you a question. Okay, so I think I'll skip this why open source. I think we all agree that open source is the powerhouse that we need if we really want to tackle climate change in general. I think it has to be that. I feel that I'll say this though. I see that there are a lot of solutions, MRB solutions out there and there are more and more machine learning and AI based solutions coming out there and one of the things is if we cannot make it available for anyone to comment and validate those models and criticize those models, people want to keep that as their say IPs and I don't know how we can build trust in nature of the solutions if we are not like open source right now. Mark Tran. Yeah, hi. Can you hear me? Yes. Yeah, it said a question. So clearly the Flint software, open source software is focused on line use, right? But it seems that there are elements that could potentially expand beyond other industries or sectors. I'm guessing that it doesn't quite cover accurate environments so I may face some other challenges. But I would imagine as far as the governance, there must be elements. The question I guess is are there elements that the software can be adapted for other industrial use cases that may not be line use at all or may just be looking at the emissions of an industry. Does this software sort of cater to those, could it be catered to those sort of other applications? And if so, how? Yeah, so yeah, so I think great question, right? So it is 100% focused currently on airflow, right? So agriculture, forestry are the land use, right? You're right, right? So that's what Flint can do right now. But at the heart of it, what Flint provides is actually a framework for building your own MRB system, right? I don't want to give this answer, yes, it can do everything, right? It can adapt. You can, you're happy to take that framework, take that foundations of how you can design this, but it would be a lot of work, right? And I'll go into a little bit more details on what happens behind the hood in Flint and then maybe that will help you kind of determine whether it can really work for the industries or not, right? So it won't be trivial for sure, right? Okay, I've been looking just briefly at the documentation, I'll take it closer later. I'll get back to you with questions. Yeah, absolutely. Ben? I'm really curious about this wide open governance slide. And I know we're kind of, we're very much focused on, I think, a lot of the same nature-based solution methodology, but we're really focused on the end customer and kind of solving the last mile logistics problem, but in a way that actually incorporates data with our farmers that can really drive towards nature-based solutions. So I'm curious if you have any use studies or you can talk a little bit more about this open governance idea, because I really like it. Yeah, so I think that the big thing for open governance was when, and I am quite new to Mojo, right? But the history there is Mojo started and Flint development started as collaboration between various different. So if you remember the very first slide came, right? There were individuals, there were governments, there were actually for-profit corporations who want to do this. And so the only way of truly keeping it democratic was through open governance, right? So I know there are a lot of open source software, which is great, but having an open governance is what enables us to be really democratic about what goes into the development of Flint. And the other key part there is the users are actually in control, right? And again, like we'll go into a slight more detail on it, but it's a framework and you can use that framework to build your own MRV systems with all the key components being provided by the software. And so users can determine like what needs to go in there. And you have to be user-centric because the need of every different user would be different, right? So yeah, big motivation behind that was having a bit more democratic process on how this will evolve, right? And also enable like all different players to come to the same table, right? So there will be representatives from the government side, there are scientists, right? And so on, right? And there's a slide which talks about exactly what the governance is, right? And Bertrand, I guess that means it was from- I have a- I see a question there from Elizabeth. I think that the answer is- so the question is for those of you who can't see, the question is so tribes in Brazil can use this to end deforestation without formal government assistance, obviously I don't want to speak on behalf of government of Brazil, but what could be done, I think one like ending deforestation is a major challenge. I don't think any one software or one one entity could solve it and it is- it is definitely a multi-stakeholder. It has to be a multi-stakeholder challenge, right? What Flint can enable is through the scenarios, right? You could run the scenarios and find say a win-win- a plausible path to a win-win solutions which could which could then nudge policy discussions and which could then motivate different stakeholders to act in the direction of avoiding and ending deforestation, right? But yeah, I think short answer is can this solve end deforestation without formal government assistance? I highly doubt it and I highly doubt any one entity or software could do that. But absolutely, I think it could be a very powerful case to be. Just to clarify that you're saying it could help create kind of a usable path that can help build momentum towards it. Right. It's going to create a bit of a simple prototype or something that shows it's viable. Right, exactly. Yeah, so I think we already discussed this, like so it can facilitate collaboration, it's a non-profit, it's no commercial interest, just very much like Hyperledge, we are part of Linux Foundation. So the reason there was simple, we wanted to focus on what we do best and focus less on things like legal and accounting and other stuff, right? And also we benefit a lot from engaging with the wider Linux Foundation community and learning about like best practices and so on. This is, I think, where the governance is. So this is how currently it is structured. So we have users, our strategy board actually consists of end users, our advisory board consists of our supporters which could, may or may not have representatives from government entities. And this is for Mojo Global, but every MRV project could have different kind of governing structure. Anybody can be on technical, that's where I sit, right? So anybody can be on technical steering committee. We have three steering committee directors, but everyone is welcome to be part of that and so on. Right, so that's the governance and people can engage in different ways. Right, so if an organization wants to just take this, use it, self run it, they are welcome to do that. If they are eager to contribute back, that's brilliant. Then we also have some collaborations with different for-profit entities who can provide say trainings or software as a service which sits on top of Flint and there's a product called Flint Pro, which is by a million group and so things like that. We all know there are different ways of commercializing and open source. But Mojo Global doesn't do that obviously. There are others who would. So now going a bit into what is the background, what happens a bit more behind the scene or in Mojo and then there's a video about I think which a lot of you would probably like a case study around how it is being used right now in one different case. Right, and the big so motivation here was that there's ground measurements and remote sensing products and all of that. But then the gap which was identified was integrating all of it together so that it becomes a system and not like one of analysis pipeline. And so that is what the big focus was. I know we have, oh wow, time flies, we have 10 minutes, I quickly go through this. At the heart of it what it does is it integrates everything. Right, and this is what is I think I would definitely want to spend some time here and the concept is time in space integration. Right, and so which is what is different than your traditional GIS models. Right, and so if you are familiar with GIS models you have maps and you can overlay those maps and you can do analysis on them different layers but what Flint does is it doesn't mean that you don't need GIS expertise. You need GIS expertise. We use this maps but what it does it does time in space calculations. Right, and so for every say a space on this all this different map and every single parcel it will it will bring in all different models and this models kind of talk with each of their ability to integrate each model at every single time step which cuts across all of these layers and that is what is I think the power of Flint which integrates over time and space but starting with time which is what most that's one of the big gaps which we have seen back and that's what the focus is. Basic systems can be complex and all I want to say is let's say you have a simple system which has a simple equation but as you start adding more complexity to it even if it's like a simple equation which is say you multiply emission factors with and get to like a gg right and so it can become very complex. Again going back to like say when it is integrated that's where you create the most value versus individual methods and measurements and this is very very subjective assessment and so I'm happy to be challenged on this but this is what well Flint what Flint does right so you have you can have all different types of data like and you have the ground data remote sensing data you do data acquisition there's like a bunch of data processing the data products but as soon as you hit reporting you need to bring all of this together you need to bring all of this together in a way which which aligns with your policy objectives which aligns with your decision making objectives and that's what Flint provides is this reporting this projections and forecasting and and integrates and generates this outputs bringing it all these different models together that's the power of Flint. The core principles are we wanted to make it flexible scalable it's data agnostic and this is a power and a challenge in itself and so like every time you go to a new country which wants to build a new MRB system the data is definitely I can get a 99% probability that it's not going to be in the same format and so you have to kind of tackle that but it's it's by design because that's what helps it make helps make Flint scalable in some sense and and this is like very high level what it does so you have you can have all this different like data inputs and this data manager which collects this data and different formats based on how we design those systems some of those formats are kind of already available but they are extensible so if you have newer formats and newer transformations you can add those transformations in it each model they're calling these modules but these are at the heart of it there are different say models right so say soil health could contain biogeochemical process models in it and then there's g a g model some of them are just say just getting started or they are just concepts for example finance module is something which we are thinking of it's not there yet but you can attach different modules to this right and and then you can take that you can integrate it you can run those those spatial time in space uh modular integration and and then create outputs in the way that's useful and yeah I do want to leave time for this demo which is having the Flint run on the cloud do you hear this and building the back end data science having the Flint run on the cloud do all of those pieces of information but the reason I wanted to show everyone this is that there's a big difference between having all of the complexity at the back and having a nice simple front end or having multiple front ends of the system depending on the user so that system you'll see some of the more sophisticated aspects of that system in the coming days where you can change databases and you can do all these really complicated things but the very same system was developed to put in place this program and there were a couple of questions about mitigation in there this is just this is a small mitigation pilot program in Australia it was designed to see if you could get some landholders interested in planting trees only small scale plantings sort of five to 200 hectare size but could you actually give them numbers that would estimate the amount of carbon they could get on their land if they planted trees so that they could make a decision on was this worthwhile entering the carbon market and trying to mitigate so it's only a test at the moment there's these six regions done around Australia but if you understand Australia in size then you see that this little box in the Burnett Mary in Queensland for example is probably half the size of France so it's actually quite big areas being simulated here it's just that Australia is obviously very very large but the tool itself was designed to be very simple so I could come into anywhere I'll choose this region in Victoria and remembering that all of the back end of this I'll close this the application program is closed but that's fine the tool still works all of the back end of this is sitting there with this sophisticated designed built system all around the flint all that happened here was another set of users came in and said we need to use that information to support this program I can come into this area here I can quickly grab my pencil click on that area and the server goes back to the wall of that model data it comes back in and it's given two or it's given three numbers it's told that that polygon that I've just drawn is 39.75 hectares of the area inside that polygon only 59% is eligible under Australia's carbon mechanisms at the moment and the reason for that is that there's trees already inside that polygon so you can't plant trees on trees and that if you were to plant those eligible areas you have 715 tons of CO2 up to 25 years that you could potentially trade so that's the reason I wanted to show that is to give an idea of yeah so that's an example I know we are out of time and I'll share slides which goes into architecture of it and what are the different modules but and I'm also happy to stay back for a few minutes if there are any questions thanks a lot for this great presentation it was a pleasure I learned a lot about flint and here there's a first question coming up so it's Tim and it's TR TR go ahead I've already asked a couple okay okay so you just represented some of the modules that your system could potentially handle like biodiversity have you worked with like the natural capital project from Stanford who has their open source models for any of the ecosystem service analysis so one I'm not aware about it but we do for example we have looked at ecosystem services which go into like different so one bio let me start with this biodiversity model like is very early right and so it's not fully integrated into the system it's not fully developed and that's why it's not integrated into the system for ecosystems modeling like the for soil for example I can tell you so there is a university in Colorado State University has like models around s luko and stuff and so we can integrate those um any I quickly having the run on the cloud I'm going to answer that question with yeah here right so this is where you you can define and bring in all different models and parameters and hyper parameters right and and I'll share these slides right but if there is a model uh which which is out there uh you could bring that as in in the system and and it can become part of the system right if you want to if there's a bespoke application which doesn't fall within the current modules you can actually even build your own model which could integrate into the system I'll jump in with my question maybe um what's in your experience what's been the usual point in the integration process where things stop working or where where has uptake where and why is uptake a flip in slower than you'd like yeah so I think the okay so the thing places where things stop working or could go wrong is that couple of things right one is you want to spend time up front in defining your objectives and policy goals right sometimes you jump into this analysis mode and start using this system and then it's not producing the results or the insights that you would really like it to right and so then then you have to go back to the drawing mode and because you're because this is an extensible system it's uh it takes time right to kind of build those things right and again right if you were to build it again right uh the other one is um there is there is a lot of data right like when you when you start thinking of like say let's say that I think there was an example here how much yeah here right so if you just say look at say your stocks and flows every time stamp 100 years monthly steps 30 pools and there could have been sub pools of carbon you're having a lot of data and so Flint makes it easy to kind of manage the data versus like we also like clear a lot of the data which is needed for models to run but not really needed for reporting right and so in designing process if we if we are not really very clear about about how we distinguish those right you will easily end up slowing the system quite a bit right and there will be a lot of computational and data storage overheads there um and and there's always there's always like when you what we would recommend is starting small and growing from that right there's always as you add more complexity to it uh there will be more more and more I'm trying to find a slide here and this is this was like a good example right so this here tier one modules and tier one database this is what I think someone asks what's available and this both are available and that's thanks to IPCC and UNFCCC we have this data globally right for for both this right but then you can start adding complexity so if you go to directly it's a highly complex system then then obviously like some places you're going to find some errors right and we recommend very high focus on qaqc actually so thank you that's fascinating okay are there any other questions are the questions in in youtube live stream no there aren't any questions um I have a further question but I know this is dependent on data but on average how what is this pixel scale resolution that you're able to accomplish yeah so uh again it depends on what you want we have done 30 meter by 30 meter so landslide data a bunch of analysis using landslide data which is that's normal I think an average parcel size you can get so um yeah we have been able to do that great nice so illegal loggers could also use this to find out where they can log next in principle um it's open source right and anyone who wants to use it can use it right so um great questions um to my knowledge we although to build this system is so complex right I mean uh usually obviously we don't know like who is everyone using it especially once it's it can be like copied and redistributed right after that so we don't know that but from our knowledge I think most of the users currently are along the lines of NDCs for governments right so Canada is using it for example Australia is using it Chile is using it and so on uh great question where are you using it in demonstrations right now so currently um there there is a is it in this there's a map but I can show you a map currently I mentioned Australia Canada Chile Belize uh there's a plan for so we we recently submitted fingers crossed a proposal and plan for adding 30 countries uh and and then go from there right so how well is it working there I mean is is is it ready for the Amazon I yeah I I think it's uh the answer is we'll find out soon um it's working well where we have so countries like Canada again have used they have been using flint for their NDC submissions right and so it follows it rigorously follows the IPCC and UNF people see like the guidelines right and and so in that sense it it's easy to verify and it's validated at least on that okay let's take one last question and then end the meeting for everyone so tiara you're raised your hand so uh Tsinghua has like one of Tsinghua University has one of the most advanced AI machine learnings for land use land classification um how are you able to keep up with um the advancements in these classification algorithms and do you have like a board that incorporates their knowledge into further design and um upkeep maintenance and integration of these yeah so I think brilliant question uh so there is a board and I am that person on the board which is looking at how we can start incorporating AI and machine learning especially like some advanced computer vision uh technologies uh into the system we are on early stages we didn't really get into that project it's uh we are running currently running a project for uh deforestation predictions and not just identification but predictions uh with uh in in Uganda and and so we want to bring in both remote sensing with on-ground data and to the focus is on building uh transfer learning models so uh you can go in places where you don't need to retrain the whole model again but uh you can find you in those models based on the regions of interest I can totally use some help in it so if uh if you know anyone who would like to volunteer their time uh on that aspect uh would be brilliant and happy to connect on that um the project is currently supported by different grants currently we are using AI for good grant from google for that uh but yeah we are definitely keeping an eye out on that and that is on the roadmap uh work in progress that sounds very exciting great to hear thank you thanks everyone for joining uh this is this was super fun and um yeah happy to answer any questions offline uh if you are interested in contributing getting engaged using Flint uh please reach out and and happy to point you to all the resources that can help you get started okay then I wonder I think what's the best way to reach out to you send me an email uh I'm also quite active on LinkedIn so Robin I can send you my contact details and you can forward it to the group uh okay I will thank you very much David it was great it was a pleasure having you I learned a lot and everyone have a great day and see you in two weeks again thanks bye