China's pulp and paper industry is one of the world's largest industrial paper systems, but its true carbon footprint has long been blurred by national and provincial averages. Now, researchers have built a plant-level carbon accounting framework that combines satellite imagery with textual industrial data to estimate emissions from 720 pulp and papermaking plants across China. The approach reveals striking differences among individual facilities, showing that emissions are concentrated in a relatively small group of high-emission plants. It also identifies which functional zones inside plants are associated with emissions and evaluates how rooftop solar deployment could help reduce emissions, offering a more precise basis for industrial decarbonization.
Traditional carbon accounting in the pulp and paper sector has mostly relied on broad statistical data, average emission factors, or energy-based calculations. These methods often miss important plant-to-plant differences in raw materials, production processes, wastewater treatment burdens, and energy structures. In a sector as diverse as China's, those blind spots can distort both inventories and policy priorities. More importantly, they make it difficult to identify which factories should be targeted first and which mitigation measures are likely to work best. Based on these challenges, deeper research is needed into plant-level, high-resolution carbon accounting for the pulp and paper industry.
Researchers from South China University of Technology reported the study (DOI: 10.1016/j.ese.2026.100682) in Environmental Science and Ecotechnology in 2026. The team developed a multimodal fusion framework that integrates high-resolution remote-sensing imagery, plant product text, and numerical data to classify factories, estimate emissions, and assess rooftop photovoltaic potential. Using this system, they produced a detailed 2022 carbon inventory for 720 Chinese pulp and papermaking plants and explored how plant-specific mitigation strategies could support the sector's low-carbon transition.
The researchers first used remote-sensing images to identify plant boundaries and distinguish functional zones such as raw-material storage, wastewater treatment, thermal power areas, and other built-up spaces. They combined remote-sensing imagery with BERT-based text classification, using imagery to distinguish two plant types and text data to separate three others that could not be identified from imagery alone. The resulting models performed strongly, reaching R² values as high as 0.96 in carbon estimation. The team estimated that China's pulp and paper industry emitted about 163.6 million tons of CO2 in 2022, with more than 60% concentrated in coastal provinces. Emissions were also highly skewed: roughly 5% of the highest-emitting plants accounted for about 43% of the sector's total. Wastewater treatment areas emerged as a consistent emission driver across plant types, highlighting a source that energy-only accounting often underestimates. The study then modeled rooftop solar deployment and found that, under the most favorable panel-length scenario, annual emissions could fall by up to 16.9 million tons of CO2, or 10.3% of sector-wide emissions.
"This work changes the scale of industrial carbon accounting from broad averages to individual factories," the study suggests in effect. By showing that emissions are unevenly distributed across plants and even across functional zones within plants, the research points toward a smarter regulatory strategy: instead of treating the whole industry as uniform, policymakers can focus first on the relatively small number of facilities where targeted interventions would have the largest payoff.
The implications go beyond papermaking. The framework offers a practical way to combine AI, satellite observation, and industrial text data to monitor complex heavy industries at fine spatial resolution. For China's pulp and paper sector, the findings support differentiated carbon-reduction policies, more efficient plant retrofits, and better prioritization of rooftop solar investment, especially in primary-fiber pulp plants with the greatest mitigation leverage. More broadly, the study provides a transferable blueprint for plant-level carbon accounting in other heterogeneous industrial sectors where hidden variation has long limited effective climate action.
