UNIST Team Advances Carbon Monitoring with Hourly GPP Forecast Model (Image supported by ChatGPT)
Ulsan, South Korea, June 2 (Korea Bizwire) – A research team at the Ulsan National Institute of Science and Technology (UNIST) has developed a cutting-edge artificial intelligence model capable of predicting how much carbon plants absorb through photosynthesis on an hourly basis—a breakthrough expected to enhance carbon monitoring and ecosystem modeling.
Led by Professor Lim Jeong-ho of UNIST’s Department of Urban and Environmental Engineering, the team created a model that estimates Gross Primary Production (GPP)—the amount of carbon absorbed by plants through photosynthesis—by training AI with high-frequency radiative and meteorological data from the Himawari-8 geostationary satellite. Unlike traditional polar-orbiting satellites, which observe the same area only a few times a day, Himawari-8 provides data at 10-minute intervals, allowing for significantly more precise temporal resolution.
GPP is a key metric used to quantify an ecosystem’s carbon uptake. According to lead author Bae Se-jung, the new model overcomes the limitations of previous satellite data by capturing minute-by-minute changes in light conditions, offering a more accurate picture of how plants respond to environmental factors throughout the day.
Crucially, the model incorporates Aerosol Optical Depth (AOD)—a satellite-derived indicator of particulate concentration such as fine dust—which influences the amount and quality of sunlight available for photosynthesis. Using an explainable AI technique known as SHAP (SHapley Additive exPlanations), the researchers identified AOD as the most influential variable during early morning and late afternoon, when the sun is low and light scattering is most pronounced.
“This model enables us to estimate hourly carbon absorption across East Asia at a 2-kilometer spatial resolution,” said Professor Lim. “It has broad applications in ecosystem carbon flow analysis, vegetation response monitoring, and light-based carbon modeling.”
The findings were published on June 1 in the international journal Remote Sensing of Environment. The study was supported by South Korea’s Ministry of Environment and the Ministry of Land, Infrastructure and Transport.
Ashley Song (ashley@koreabizwire.com)
