Machine Learning for Peatland Ground Water Level (GWL) Prediction via IoT System

Abstract

Peatland poses a severe environmental threat due to its potential for massive carbon emission during fires. Conventional Ground Water Level (GWL) monitoring in peatlands is labor-intensive and lacks real-time data, hindering effective management. To address this, this paper proposed an IoT system with neural network-based GWL prediction for real-time monitoring. By using atmospheric parameters, the neural network predicts GWL, allowing extra time for the responsible party to take the appropriate action to reduce the fire risk in peatland. The proposed neural network demonstrates promising results, with a Root Mean Square Error (RMSE) between 3.554 and 4.920, ensuring 99% accuracy within 14.760 mm range of the actual GWL. This finding underscores the novel approach of integrating IoT and neural networks for peatland GWL prediction, offering a significant advancement in real-time monitoring and fire risk mitigation strategies. The novelty lies in its capability to predict real-time GWL even in areas lacking the resources for conventional monitoring, using simple meteorological parameters.

Keywords

• Peatland
• IoT system
• machine learning
• neural network
• ASEAN
• transboundary haze