SDG 1: No Poverty
SDG 2: Zero Hunger
SDG 3: Good Health and Well-being
SDG 4: Quality Education
SDG 5: Gender Equality
SDG 8: Decent Work and Economic Growth
SDG 13: Climate Action
SDG 14: Life Below Water
SDG 15: Life on Land
2. Project Details
Company or Institution
Atoma Solutions GmbH
Vlinder – Mangrove Carbon Verification System
General description of the AI solution
"Vlinder – Mangrove Carbon Verification System" is a project to explore carbon sequestration and deposition processes in mangrove forests in a machine-learned capture model scalable for CO2 certificate trading.
In this context, new methods in data aggregation (satellite data, LiDAR data from drone images, soil sampling), and modelling (convolutional neural networks, computer vision) are used to construct a more accurate method of measuring CO2 content in mangrove forests.
Although there are significant scientific gaps in this area, mangroves were chosen as the focus of development because of their superior CO2 storage capacity as well as the overall ecological added value. The additional data is collected on our own, together with local aid organizations and rural communities, which are included in the project to achieve long-term sustainable solutions.
The method can be applied in near real-time to continuously monitor the carbon content in forests. This is especially important due to potential floods or other natural or man-made reductions in the mangrove forest.
In a hackathon the model was already successfully implemented, with a higher accuracy than traditional methods. Additional parameter tuning is planned and the additional data from soil and ground truthing potentially further increases the accuracy.
The near real-time measurement approach and the scalable aggregation model additionally contribute to promote new green projects to sustainably advance the green economy.
Overall, the AI solution based on new, self-collected data and fine-tuned convolutional neural networks for computer vision will greatly improve the accuracy of measuring carbon content in Mangrove forests (and potential other forests). This increased accuracy will enable funding via carbon certificates and further environmental projects as the far reaching environmental and economic value of mangrove forests can be predicted in a more comprehensive way.
Atoma Solutions GmbH
Excellence and Scientific Quality: Please detail the improvements made by the nominee or the nominees’ team or yourself if your applying for the award, and why they have been a success.
The type of AI is a convolutional neural network, used for computer vision. The model will be applied on satellite and drone images as well as LiDAR data collected via drone flights.
The complexity of mangroves arises from the fact that the carbon sequestration process takes place at different reaction levels (Above Ground Biomass (AGB) in the canopy, Below Ground Biomass (BGB) in the root system and in the soil (Soil Carbon). While there is already substantial research on ABG, the BGB and soil carbon measurements lack crucial research insights to develop a comprehensive model. That is where the AI solution will excel and outperform current methods, as well as add innovation to the research field. In cooperation with the research institution International Institute for Applied Systems Analysis (IIASA) in Lower Austria, a near real-time specific remote sensing model will be scientifically developed for the first time. IIASA is an expert on ground-truthing and AGB estimations in general, while our team adds the knowledge of AI modelling and data processing capabilities.
The technology readiness level (TRL) is TRL 6 and best described with our hackathon, where the model was already successfully implemented on a small mangrove forest in Myanmar, with a higher accuracy than traditional methods, focusing on Random Forest methods. Additional parameter tuning is planned and the additional data from soil and ground truthing from local groups potentially further increases the accuracy. The results were presented, and our researchers are frequently part of conferences and with media coverage.
In general, the values of the company lead to open-source AI development and algorithm solutions to enable other carbon verification systems to improve the accuracy. The data will be purchased and owned by the company.
Scaling of impact to SDGs: Please detail how many citizens/communities and/or researchers/businesses this has had or can have a positive impact on, including particular groups where applicable and to what extent.
More accurate carbon content measurements lead to improvement for all involved stakeholders of projects. Vlinder already has projects implemented in Myanmar, Kenya and Sri Lanka. These interventions lead to measurable progress of the following SDGs:
1 – Total number of people expected to have improved livelihoods or income generated because of project activities -> 16,000 people in 3,800 families.
2 – The project provides training on how to improve agricultural productivity which will lead to better food security.
3 – The entire community is expected to achieve improved well-being because of project activities as healthy forests improve the quality of life of the people who depend on them.
4- Total number of community members with improved skills and knowledge resulting from training as part of project activities is 4,000. Improved income also has a positive effect on the level of children's education.
5- At least 50% of people who would have improved livelihoods or income generated through the project are women. This set up helps to improve equality and power dynamics, while providing opportunities to women.
8 – One project will create 200 full-time jobs and will provide training for young people to improve their work prospects in the future.
13- The project is preserving forests that store carbon dioxide with a positive effect on the global climate.
14 & 15 – The project encourages sustainable farming and land stewardship as well fish habitat (through mangroves).
The projection of global impact is clear: every reforestation project wants to have control over and be able to easily double check their carbon content to prove environmental value and stimulate investment. This affects every country with mangrove reforestation potential. Furthermore, we plan to contribute to further development of AI for SDGs by providing the AI solution to projects in other ecosystems.
Scaling of AI solution: Please detail what proof of concept or implementations can you show now in terms of its efficacy and how the solution can be scaled to provide a global impact ad how realistic that scaling is.
Our evidence for impact is: (i) our own, already implemented Myanmar and Kenya projects as case studies, (ii) mentions in Carbon Pulse (renowned publisher in the carbon markets) and (iii) our hackathon with the successful implementation of the AI solution.
The scalability of the solution is based on the potential to use the AI solution in other mangrove forest areas. This amounts to around 812 000 hectare and a potential restoration value of 420 million Mg of CO2. Each restoration project is improved by an accurate measurement method of CO2 content.
Additionally, other companies will also start using AI because it is more accurate and cheaper than the current state of the art, thanks to remote sensing technologies. If the companies that manage other ecosystems (e.g.tropical and boreal forests) adopt our solution, spill over effects will happen and overall ecosystem management can be improved with the help of AI.
A particular part of the project that involves people and democratizes the technology is our citizen science component. The ground truthing of our model is performed by people from local communities. All a person needs to participate is a smartphone with a camera, and even very remote villages that are located in our project areas have those available. Our research partner, IIASA has already created tools and an app to enable such projects. We are aware that the bandwidth poses a challenge, but we are already working to store the information locally and then send it batchwise.
Our early adopters are carbon brokers like the Compensate NGO headquartered in Finland, that adds a layer of transparency to their project and offers a strategic partnership which helps us to reach many clients.
Our company is ideologically open source, which includes even publicizing our parameter tuning, while the data stays our own.
Ethical aspect: Please detail the way the solution addresses any of the main ethical aspects, including trustworthiness, bias, gender issues, etc.
The ethical considerations and implications of our AI are manifold. While the AI itself does not face the issue of bias towards humans as we only measure carbon content in trees, the impact of the solution on trustworthiness of restoration projects and enabling gender equality is crucial. By including local communities in planting and protecting the mangroves and using half of the profit to finance these measures, we ensure that the forest is not simply burned down or logged after leaving the site.
The data collection and AI developments are conducted in a lawful manner, complying with all applicable regulations in the region and at our HQ in Austria. After training the model, we can provide a trustworthy and robust way of measuring the carbon content of our own and other restoration projects around the globe.
Furthermore, with the Citizen Science component we increase the inclusiveness of the solution, as the data collection is mostly done by local communities, which proves to be often their only source of income. The digitalization also enables many citizens with smartphones (only way of banking in some regions) to have a source of income. This is a central part of our project to include local communities.
Overall, the combination of the above promotes a long-term solution, which is in the best interest of us as project initiator and the local aid organisations and communities as project executors.