Scholarlink

This study aims to predict PM2.5 concentrations due to forest and peatland fires in Riau Province, Indonesia, by utilizing satellite, meteorological, and land cover data. Riau Province was chosen because it has the largest peatland in Sumatra and often experiences forest fires that cause cross-border haze. The ensemble learning model used involves a combination of Gradient Boosting Machine (GBM), XGBoost, and Neural Network (NN). Hyperparameter tuning and feature selection using XGBoost were performed to improve the accuracy of the model. The results showed that the optimized model performed best with an R2 of 0.89, MAE of 2.74, and RMSE of 3.70. The model was able to explain 89% of the data variability and had a low prediction error rate. This is the first study to use this approach in Riau Province, and the results are expected to contribute positively to environmental crisis management and public health policy in Indonesia.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of Ensemble Learning with Spatio-temporal Stacking Regressor for PM2.5 Prediction from Forest Ecosystem Fire Smoke

  • 章节

  • 发表日期：

    2026年

查看全文
  • Mitra Unik,
  • Imas Sukaesih Sitanggang,
  • Lailan Syaufina,
  • I. Nengah Surati Jaya

摘要

This study aims to predict PM2.5 concentrations due to forest and peatland fires in Riau Province, Indonesia, by utilizing satellite, meteorological, and land cover data. Riau Province was chosen because it has the largest peatland in Sumatra and often experiences forest fires that cause cross-border haze. The ensemble learning model used involves a combination of Gradient Boosting Machine (GBM), XGBoost, and Neural Network (NN). Hyperparameter tuning and feature selection using XGBoost were performed to improve the accuracy of the model. The results showed that the optimized model performed best with an R2 of 0.89, MAE of 2.74, and RMSE of 3.70. The model was able to explain 89% of the data variability and had a low prediction error rate. This is the first study to use this approach in Riau Province, and the results are expected to contribute positively to environmental crisis management and public health policy in Indonesia.