Rostov Region, October 14, 2025, DON24.RU. What weather should you prepare for this winter? What awaits us in the usual tomorrow – a stubbornly climbing thermometer or a notable cold snap? When will the low water end? Researchers at the Department of Oceanography of the Southern Federal University, engaged in truly unique research throughout the Russian Federation, are trying to find answers to these and many other questions. We are talking about perfecting the climate model for long-term determination of weather trends in Rostov and its surroundings using artificial intelligence. Now these works have become part of the large SFU grant project “Development and implementation of a method for creating a digital copy of soil based on artificial intelligence and Big Data technology”, which will be completed in 2030. A journalist from the Molot newspaper learned about some provisional results. Scoring on AI A few years ago, the news about the beginning of development of a model designed to determine weather patterns in Rostov and its surroundings with a foresight of 20–30 years (!) became almost a sensation. To help themselves, scientists have called for neural networks with hybrid architecture – Convolutional Long Short Term Memory, gradually training them. ERA-5 reanalysis data from the European Center for Medium-Range Weather Forecasts are also used, which are relevant to weather over the Don River. “We achieved good results in terms of rainfall. Our program predicted heavy rainfall last year, but for 2025 it predicted very little rainfall. And that happened,” the climatologist explained. O. Head of the Department of Oceanography, Candidate of Geographical Sciences Alexander Ioshpa. “In general, we have learned to determine quite accurately the air temperature and precipitation in Rostov and the surrounding area two weeks in advance.” The immediate goal is to determine weather parameters by year. But scientists emphasize that their task is not to replace hydrometeorologists and determine the exact temperature (to a tenth of a degree) every day, but to determine the main characteristics of how exactly the climate will change over the next dozen years, whether certain parameters will differ from the average monthly and average annual climate norms. At its peak For the digital twin of the Don lands, data on air temperature and precipitation obtained from January 2001 to the end of 2024 at five weather stations in different parts of the region will be loaded into the program. “These are stations in Rostov, Tsimlyansk, Chertkov, Remontny and the village of Gigant, Salsky district,” explains Denis Krivoguz, leading researcher at the Department of Oceanography at the Institute of Earth Sciences of the Southern Federal University, Candidate of Geographic Sciences. With climate change, land cover and vegetation also change. We also analyze whether the parameters change synchronously or not. And the innovation of the project lies in its scale: we are bringing together a digital giant, because the situation is being studied not in one sector or one agricultural enterprise but across an entire region.” In terms of temperature, the warmest year in the Don since the beginning of meteorological observations was last year, 2024. In it, the average annual temperature in Rostov was +12.6 ° C, for comparison: in the distant 1896 and 1933 it was +7.8 ° C. But it is too early to conclude that the thermometer will only continue to rise. “The average temperature in summer 2024 in Rostov is +25.5 °C, and in 2025 it is lower, +24.3 °C,” explains Alexander Ioshpa. – In general, our program shows that in Rostov we have reached a certain warming peak. According to our version, we will stay at this level for 10–15 years, but then the weather may get colder. However, for the model to function fully, there is not enough data on the circulation of air masses; We will collect it.” As for the approaching winter, the program shows it will be colder than last year, Alexander Ioshpa said. Hell in the Concrete Jungle Don scientists also note that climate suffering is partly due to… urban planning decisions. This year, SFU received results from the research project “Intensity Index of the Urban Heat Island in Rostov-on-Don”. Scientists used Landsat satellite data for 2016, 2020 and 2024, as well as AI algorithms. We studied several areas of the Leventsovsky residential district in Rostov. If we get rid of complicated terminology, it turns out that areas built with houses and roofed with bricks and asphalt will warm up much more than areas with trees and water. As a result, urban areas warm up to +35–50 °C, water surfaces remain within +28–35 °C, where Many trees, thermometer is +28–40 °C. “Cities need trees, parks, fountains and artificial ponds, and ideally combine water and plant environments,” says Denis Krivoguz. By the way, Alexander Ioshpa emphasized that the reasons for water shortage are complex. “Obviously winters with little snow have an impact, but the size of growing cities is also significant,” he said. – In fact, when there are many people living in rural areas, for water bodies will be gentler: there is no point in consuming huge amounts of water. Insufficient demand for major water consumers is also unacceptable. Two years ago, my colleague and I met a representative of the agricultural complex. He admitted: “After they started installing agricultural meteorological stations to predict weather developments, we saved 30% of water.” If we want to solve the problem of low water levels, This approach is needed everywhere! In addition, the springs must not be clogged with waste or waste because it will prevent water from flowing into the reservoir.”
