Scholarlink

<p>Subnational solar energy potential forecasting is crucial for the sustainable energy transition in developing countries, particularly during periods of heightened climate uncertainty. This research presents a hybrid big data technique in this paper that evaluates the feasibility of temperature-based solar energy in Rajshahi and Ishwardi, two underserved but very solar-rich districts of Bangladesh, using both deep learning and statistics. This study analyzed 40 years (1980–2020) of historical climate data provided by the Bangladesh Meteorological Department (BMD), together with NASA-POWER satellite variables (irradiance, clearness index, albedo, and cloud cover), to evaluate linear trends and nonlinear seasonal fluctuations. Using an LSTM neural network and linear regression, the mean temperature, precipitation, and humidity were forecasted. Predicting errors were decreased by more than 70%, and the LSTM model defeated the linear models by a margin of more than 70%, with RMSEs of 0.25 and 0.22&#xa0;°C in Rajshahi and Ishwardi, respectively. Spatial diagnostics show that Rajshahi is a better location for photovoltaic system deployment because of its consistently higher level of sun irradiation and lower levels of cloud interference. The study provides a scalable method for energy planning for climate resilience and demonstrates the viability of the AI’s predictions in contexts with little data. The findings, based on SDGs 7 (Affordable and Clean Energy) and 13 (Climate Action), will inform country-specific laws, highlighting the importance of geospatial analysis and predictive climatology in ensuring equitable and sustainable energy access.</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

Temperature-based solar energy forecasting: a big data analysis for sustainable energy planning in Ishwardi and Rajshahi region of Bangladesh

  • 期刊论文

  • 发表日期：

    2025年9月27日

查看全文
  • Hasan Ahamed Alif

摘要

Subnational solar energy potential forecasting is crucial for the sustainable energy transition in developing countries, particularly during periods of heightened climate uncertainty. This research presents a hybrid big data technique in this paper that evaluates the feasibility of temperature-based solar energy in Rajshahi and Ishwardi, two underserved but very solar-rich districts of Bangladesh, using both deep learning and statistics. This study analyzed 40 years (1980–2020) of historical climate data provided by the Bangladesh Meteorological Department (BMD), together with NASA-POWER satellite variables (irradiance, clearness index, albedo, and cloud cover), to evaluate linear trends and nonlinear seasonal fluctuations. Using an LSTM neural network and linear regression, the mean temperature, precipitation, and humidity were forecasted. Predicting errors were decreased by more than 70%, and the LSTM model defeated the linear models by a margin of more than 70%, with RMSEs of 0.25 and 0.22 °C in Rajshahi and Ishwardi, respectively. Spatial diagnostics show that Rajshahi is a better location for photovoltaic system deployment because of its consistently higher level of sun irradiation and lower levels of cloud interference. The study provides a scalable method for energy planning for climate resilience and demonstrates the viability of the AI’s predictions in contexts with little data. The findings, based on SDGs 7 (Affordable and Clean Energy) and 13 (Climate Action), will inform country-specific laws, highlighting the importance of geospatial analysis and predictive climatology in ensuring equitable and sustainable energy access.