MAI will teach artificial intelligence to assess land productivity.

Researchers at the Moscow Aviation Institute are developing the world's first big data database in the field of soil science. The project encompasses satellite imagery of arable land across various spectral ranges collected over the past four years. Experts have already begun working on an AI-based program that will identify areas of increased soil fertility.

2024-10-23 07:31:30https://naked-science.ru/article/column/iskusstvennyj-intellekt-o

The starting experimental site for the implementation of this unique project will be the Kurkinsky district of the Tula region. Throughout 2024, MAI employees conducted active work on the automated collection of satellite images and additional data, which will serve as the foundation for the project. This data set will enable artificial intelligence to identify fertile soils.

Through automated image analysis, AI will be able to determine areas with the highest and lowest productivity. This will allow farmers to optimally distribute fertilizers, focusing on areas with low nutrient content.

According to the project leader, candidate of biological sciences and associate professor of the "Ecology, Life Support Systems and Life Safety" department at MAI, Sergey Ogorodnikov, the use of machine learning methods allows for the classification of soils and the identification of hidden dependencies between their physicochemical and biological characteristics.

“Russian President Vladimir Putin has tasked us with increasing agricultural production by at least 25 percent by 2030 compared to 2021. The country's leadership has set the goal of ensuring food security. To increase yield and productivity, it is critically important to rationally and effectively apply fertilizers and understand how soil-ecological conditions change within fields. This method helps clarify the relationship between the spectral characteristics of the soil and the vegetation on it. Soils have the ability to absorb and reflect various types of light, which is visible in infrared images,” notes Sergey Ogorodnikov.

Right now, Tula farmers are showing great interest in the project. While working in the fields, they have started to rely on calculations and satellite images collected by MAI specialists.

“We are receiving real orders from agribusiness holdings; this year we surveyed 60 thousand hectares. For the research, we prepared a sampling grid, dividing fields into squares of 10 hectares, taking into account the terrain and soil productivity. Without artificial intelligence, it would have been impossible to solve such a task manually,” says the scientist.

The commercialization of research results has begun thanks to the "Umnik" grant from the Innovation Support Fund. Currently, the team has submitted an application for the next stage of the competition - "Start-AI".

As a result of the work, two patents have been obtained: for the agro-soil-ecological database of the Tula region and for a program that models the distribution of pollutants in the soil, taking into account the terrain.

“Automated image analysis will allow us to identify productivity zones in the fields. As a result, farmers will be able to optimize fertilizer application by redistributing them to nutrient-deficient areas. These management and reclamation solutions can enhance yield and ensure sustainable territorial development aimed at combating soil degradation,” added Sergey Ogorodnikov.

This material was prepared with the support of the Ministry of Education and Science of Russia.