Scholarlink

<p>This study addresses a critical gap in climate and public health planning for Somalia by identifying the most accurate forecasting model for its annual methane (CH4) emissions. Given Somalia’s vulnerability to environmental change and the significant contribution of its agricultural sector (particularly livestock) to CH4 emissions, robust projections are essential for effective mitigation. Utilizing annual time-series data from 1960 to 2023, the research rigorously compared fifteen models, including six individual approaches (ARIMA, ETS, TBATS, Theta, ARFIMA, NNAR) and nine hybrid combinations. Data stationarity was ensured through first-order differencing. Model performance was evaluated using four distinct error metrics: Theil’s U, Mean Absolute Percentage Error (MAPE), Symmetric Mean Absolute Percentage Error (SMAPE), and Mean Absolute Scaled Error (MASE). While the ARIMA (0,1,1) model with drift emerged as the best-performing individual model, the hybrid ARIMA-Theta model significantly outperformed all other methods. It demonstrated superior predictive accuracy across all evaluation metrics (Theil’s U: 0.7471; MAPE: 0.4835; SMAPE: 0.0049; MASE: 0.2334). Forecasts from this optimal hybrid model project a continued, substantial increase in Somalia’s CH4 emissions, anticipated to reach approximately 27.42&#xa0;million metric tons by 2033. These findings underscore the superior effectiveness of hybrid models for forecasting complex environmental data like CH4 emissions. The projected rise in emissions highlights an urgent need for national intervention, providing crucial, data-driven insights for Somali policymakers to craft targeted mitigation strategies in agriculture and waste management, integrate climate action into public health planning, and fulfill national and international environmental commitments in a climate-affected and vulnerable nation.</p> Graphical Abstract <p></p> <p>This study pioneers the first comprehensive forecasting of Methane (CH4) emissions in Somalia, a nation particularly vulnerable to climate change impacts, by analyzing 64 years of annual data from 1960 to 2023. As depicted in the graphical abstract, the research initiates with raw data for Somalia (represented by the country’s outline and a droplet for historical data from Our World in Data [OWID]) which then undergoes rigorous analyses, including ADF/PP tests and a comparison of 6 single and 9 hybrid time series models. The central finding, highlighted in the “Model” section, demonstrates that the ARIMA-Theta (best) hybrid model consistently outperformed all others in predictive accuracy. The “Results” section quantitatively supports this, showcasing the ARIMA-Theta’s superior performance across key metrics such as MASE (0.2334), MAPE (0.4835), Theil’s U (0.7471), and SMAPE (0.0049), notably outperforming the best single model MASE (0.3165). This robust model projects a significant increase in Somalia’s CH4 emissions to 27.42 million metric tons by 2033 (visually represented with an upward arrow and the projected value). The “Results” section further underscores the practical implications, with icons representing livestock (cow) and waste management (trash bin with leaf) to indicate the primary sources of these emissions. This research provides crucial data-driven insights for Somali policymakers, enabling the development of targeted mitigation strategies in agriculture and waste management to enhance climate resilience and public health.</p>

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

Forecasting Methane Emissions in Somalia: A Hybrid Time-Series Approach for Climate and Public Health Insights

  • 期刊论文

  • 发表日期：

    2025年12月22日

查看全文
  • Mohamed Abdi Abdullahi,
  • Abdisalam Hassan Muse,
  • Hamse Abdi Aden,
  • Saralees Nadarajah

摘要

This study addresses a critical gap in climate and public health planning for Somalia by identifying the most accurate forecasting model for its annual methane (CH4) emissions. Given Somalia’s vulnerability to environmental change and the significant contribution of its agricultural sector (particularly livestock) to CH4 emissions, robust projections are essential for effective mitigation. Utilizing annual time-series data from 1960 to 2023, the research rigorously compared fifteen models, including six individual approaches (ARIMA, ETS, TBATS, Theta, ARFIMA, NNAR) and nine hybrid combinations. Data stationarity was ensured through first-order differencing. Model performance was evaluated using four distinct error metrics: Theil’s U, Mean Absolute Percentage Error (MAPE), Symmetric Mean Absolute Percentage Error (SMAPE), and Mean Absolute Scaled Error (MASE). While the ARIMA (0,1,1) model with drift emerged as the best-performing individual model, the hybrid ARIMA-Theta model significantly outperformed all other methods. It demonstrated superior predictive accuracy across all evaluation metrics (Theil’s U: 0.7471; MAPE: 0.4835; SMAPE: 0.0049; MASE: 0.2334). Forecasts from this optimal hybrid model project a continued, substantial increase in Somalia’s CH4 emissions, anticipated to reach approximately 27.42 million metric tons by 2033. These findings underscore the superior effectiveness of hybrid models for forecasting complex environmental data like CH4 emissions. The projected rise in emissions highlights an urgent need for national intervention, providing crucial, data-driven insights for Somali policymakers to craft targeted mitigation strategies in agriculture and waste management, integrate climate action into public health planning, and fulfill national and international environmental commitments in a climate-affected and vulnerable nation.

Graphical Abstract

This study pioneers the first comprehensive forecasting of Methane (CH4) emissions in Somalia, a nation particularly vulnerable to climate change impacts, by analyzing 64 years of annual data from 1960 to 2023. As depicted in the graphical abstract, the research initiates with raw data for Somalia (represented by the country’s outline and a droplet for historical data from Our World in Data [OWID]) which then undergoes rigorous analyses, including ADF/PP tests and a comparison of 6 single and 9 hybrid time series models. The central finding, highlighted in the “Model” section, demonstrates that the ARIMA-Theta (best) hybrid model consistently outperformed all others in predictive accuracy. The “Results” section quantitatively supports this, showcasing the ARIMA-Theta’s superior performance across key metrics such as MASE (0.2334), MAPE (0.4835), Theil’s U (0.7471), and SMAPE (0.0049), notably outperforming the best single model MASE (0.3165). This robust model projects a significant increase in Somalia’s CH4 emissions to 27.42 million metric tons by 2033 (visually represented with an upward arrow and the projected value). The “Results” section further underscores the practical implications, with icons representing livestock (cow) and waste management (trash bin with leaf) to indicate the primary sources of these emissions. This research provides crucial data-driven insights for Somali policymakers, enabling the development of targeted mitigation strategies in agriculture and waste management to enhance climate resilience and public health.