AI plays a part in sustainable forestry
Preparation for future risksNatural disaster risk predictionFull custom AI development
Clients
NeXT FOREST Inc.,1 a joint venture of Sumitomo Forestry and IHI Corporation, has collaborated with Recursive Inc., to build an early AI model (AI hydraulic model) for tropical peatland management.2; With the introduction of this technology, it will become possible to carry out groundwater level predictions with the help of AI, something that only experienced engineers were previously able to do, based on the topographic maps and actual groundwater level measurement data that Sumitomo Forestry has collected over a span of 10 years.
■ Background of Development
Sumitomo Forestry and IHI Corporation are jointly developing a Hydraulic Model3 that simulates groundwater levels to effectively manage tropical peatlands. The management of tropical peatlands is crucial for addressing climate change issues, preventing peatland fires, and conserving biodiversity.
Hydraulic models require a substantial amount of time and effort to gather and analyze diverse data, including precipitation, plant transpiration, elevation, slope, and weather conditions. That being said, due to the pressing nature of such issues, there is a necessity to adopt quicker and more efficient solutions. To address this challenge, Sumitomo Forestry and IHI Corporation are working on the practical application of AI technology to expedite the early deployment of groundwater level management techniques.
The development of the said AI technology is being done in collaboration with Recursive Inc., which was established in Japan in 2020 by Dr. Tiago Ramalho, a senior researcher at DeepMind, an AI development company under Alphabet, Google's parent company. This collaboration aims to leverage Dr. Ramalho’s expertise to advance the practical application of AI in the early deployment of groundwater level management technologies.
The AI model is a unique creation that combines AI-based machine learning using groundwater level data, which has been measured over more than 10 years in tropical peatlands in Indonesia managed by Sumitomo Forestry, as training data4 and physics modeling5. With this technology, it becomes possible to predict groundwater levels for the entire designated area up to 7 days in advance. The predictions are made by leveraging various inputs, including topographic maps, measured groundwater levels at observation points, and actual precipitation data.
In the future, there are plans to enhance the model so that each client can utilize the model to run simulations on the peatlands they manage, which will enable them to assess the impact of different channel operations and dam placements on maintaining proper groundwater levels.
1Established in February 2023 to provide consulting services for the appropriate management of tropical peatlands.
2Peat deposits are formed when the remains of plants do not decompose in water. When the groundwater level drops and the land dries out, the peat, which contains a lot of carbon, not only decomposes and disappears but also becomes highly flammable, making groundwater level management extremely important.
3A physics model based on the principles of hydraulics. Hydraulic Model is used to replicate the water cycle in nature, such as the flow of water.
4Input data and corresponding correct data that is used when training AI.
5A diagram, etc., that describes an object in a detailed representation that closely resembles its substance. A representation of specifications or a blueprint.
＜Flowchart of Physics-Based AI Model Architecture and Image of Groundwater Level prediction＞
Sumitomo Forestry leverages a decade worth of accumulated data in its forest operations in West Kalimantan, Indonesia.
The AI model is fed with a range of data sources, including topographic details like land elevation and slope, rainfall statistics, and information on waterways.
The AI model, that integrates an AI machine learning model with a physics-based model, forecasts groundwater levels several days into the future. The AI model targets groundwater level data when conducting its machine learning process.
The system is designed to issue a notification when the AI model predicts that groundwater levels in the forthcoming days may approach levels associated with potential fire hazards.
■Project Background
Tropical peatlands, land areas marked by an accumulation of undecomposed organic matter from dead plants, are significant reservoirs of water and carbon. Spanning regions like Indonesia, the Congo Basin, and the Amazon, these peatlands cover over 82 million hectares globally (approximately twice the total land area of Japan) and are believed to hold at least 89 billion tons of carbon, nearly tenfold the worldwide carbon emissions in 2017.
Unregulated destruction of these peatlands through logging or slash-and-burn activities can lead to the decomposition of this organic matter, consequently releasing massive amounts of carbon dioxide into the atmosphere. This could disrupt the essential hydrological cycle of absorbing rainwater underground and releasing it back into the atmosphere through evapotranspiration, potentially resulting in extreme weather events such as droughts and floods.
Proper management of the groundwater levels in tropical peatlands is of utmost importance. If levels drop excessively, drying out can occur, turning the peatlands into highly flammable areas prone to forest fires. Conversely, overly high groundwater levels can inhibit tree growth, underscoring the necessity of continually maintaining an optimal balance.
Sumitomo Forestry and IHI recognize the crucial necessity of conserving and appropriately managing tropical peatlands. Such actions are imperative for maintaining a stable environment for human life, including the achievement of carbon neutrality and the preservation of biodiversity and water cycles. To this end, IHI and Sumitomo Forestry are developing consulting services aimed at promoting worldwide implementation of tropical peatland management technology. This service merges Sumitomo Forestry's established high-precision land observation system for tropical peatlands with IHI's expertise in drone and satellite data utilization and weather observation and prediction technology, focusing on the aerospace sector.
Recursive is a service provider of AI solutions for building a sustainable future. We provide AI system development and consultation services by combining our knowledge of diverse industries such as environment, energy, medical, pharmaceutical, food, and retail with our advanced technological capabilities and expertise in sustainability business. Our unparalleled professionals lead the way in creating a new society with world-class technology in order to leave a better global environment and society for future generations.
