Sber and Skoltech: Carbon Balance Software

Main article: Data mining Intelligent data analysis

2022: As part of the Sber and Skoltech solution for assessing the carbon balance

On February 8, 2022, Sberbank announced that in the framework of cooperation between Sber and the Skolkovsky Institute of Science and Technology (Skoltech), software (software) will be developed to estimate the carbon balance from remote sensing data. This software will allow comprehensive analysis to make effective decisions in the field of decarbonization and develop strategies for reducing the carbon footprint, and the use of artificial intelligence ( AI) will help to improve the accuracy of the results and accelerate the analytics process in large territories.

Sber is an industrial partner of the Research Center in the field of artificial intelligence based on Skoltech, created at the end of 2021 with the support of a federal grant. The activity of the center is aimed both at the development of fundamental AI technologies, and at the development of applied AI tools for monitoring, forecasting and optimizing ESG risks, as well as solving problems in the field of decarbonization.

As part of the project, in 2021, the Skoltech team, together with the scientific partners of the project (IKI RAS, IO RAS named after Shirshov, IGE named after Israel), collected a set of up-to-date data on sources and sinks of carbonaceous greenhouse gases and developed a software concept. This is the first step to further develop and adapt AI methods to the subject area of the carbon budget. At the same time, a list of pilot regions for testing and piloting the software under development was determined.

The quality of training datasets is important to ensure the accuracy and reproducibility of AI algorithms, so data were collected for pilot regions describing the state of forest areas with a total area of ​ ​ more than 6.5 million hectares based on remote sensing of the Earth and statistics from 2010 to 2020.

Based on these data, the total carbon cycle process biochemical models for forest areas estimated carbon stocks and flows. To assess the carbon balance of forest areas, hybrid approaches based on carbon cycle modelling have been developed, as well as deep neural networks have been used to analyze Earth remote sensing data.

The development and testing of software is planned to be completed by the end of 2024. The project was coordinated by the Sber Research and Innovation Department.

"Achieving carbon neutrality today is among the priorities not only of Sber and Russia, but also of the entire planet. Climate change must be slowed down before it leads to irreversible consequences. To monitor progress in this direction, we need appropriate software, which we began to develop together with Skoltech. I am sure that the use of artificial intelligence will help make this tool as accurate and relevant as possible to the tasks set. As you know, you can effectively control only those processes that can be clearly measured, "- says Alexander Vedyakhin, First Deputy Chairman of the Board of Sberbank.

"Moving the world economy towards a green agenda can lead our country to a deep crisis if we do not learn, at least, to consider our carbon balances as methods recognized by the international scientific community. There is a basis for this in Russia. Now we need to prove not only theoretically, but also experimentally that these methods really work. And I am sure that together with Sber we will fulfill this task, "- notes Alexander Kuleshov, rector of Skoltech.