According to research, global food demand is expected to increase by 110% by 2050, while 40% of croplands and pastures are currently under threat from climate change and other factors. The study used open data and AI to predict that agricultural land will move towards northern territories in the next 25 years. Valery Shevchenko, a research engineer at Skoltech's Applied AI Center in Russia and the first author of the work explained these data sources in more detail.

The research was conducted in three stages: data collection and preprocessing, machine learning model training, and results evaluation. The study focused on Eastern Europe and Northern Asia and used openly available data sources such as ERA5 for climate analysis and CMIP models for predicting climate change until 2100. Three data sets were analyzed for different climate change scenarios: a sustainable, low-emission future, a moderate emission trajectory, and a high fossil fuel dependency scenario. The team also integrated global food security data at 1 km x 1 km resolution to conduct an in-depth study of the irrigation of arable land.

The researchers developed a model that accurately predicts the current state using the data collected from CMIP models. This model can project what may happen in 2050. However, it is essential to consider various factors such as land types and soil erosion when making these predictions. While the accuracy of these projections cannot be guaranteed, the goal is to raise awareness and encourage proactive strategies for the future.

Shevchenko predicts that in the next 25 years, there will be an increase in arable land moving northward, which may pose potential risks due to the need for increased irrigation in currently exploited regions. These findings align with recommendations from the Intergovernmental Panel on Climate Change for detailed regional assessments to adapt to climate variability and maintain food supplies.