Forecasting Oak Diameter Using an LSTM Neural Network in the Missouri State of USA

Accurately predicting tree growth is crucial for sustainable forest management and addressing climate-related impacts on ecosystems. This study explores the use of Long Short-Term Memory (LSTM) neural networks to forecast the Diameter at Breast Height (DIA) for white oak and red oak trees in Missouri, USA, which is a key indicator of tree health, growth, and timber value. Using historical data from 1999 to 2020 for training and 2021 data for validation, the model demonstrated its capability to capture growth patterns effectively. Results revealed that the LSTM model achieved higher predictive accuracy for white oak, attributed to its more consistent growth trends, whereas red oak predictions were influenced by greater sensitivity to environmental variability. These findings emphasize the potential of LSTM models in forestry analytics providing detailed insights into species-specific growth dynamics and supporting enhanced forest resource monitoring and management. The study contributes to advancing machine learning applications in forestry by showcasing how LSTM networks can address challenges in predicting tree growth across diverse species and environmental conditions. This research highlights the transformative role of data-driven models in fostering sustainable forest management practices and improving decision-making processes in the face of ecological and climate uncertainties.