Scholarlink

Weather forecasting in countries like Bangladesh poses unique challenges, given its diverse geographical features. The region experiences varying weather patterns influenced by factors such as monsoons, river systems, and coastal influences. The intricacies of high temperatures, sudden atmospheric changes, and complex circulations demand advanced forecasting methods. The research aimed to address the challenges in weather prediction for 8 cities in Bangladesh. The research used 3 deep learning models: CNN(Conv1D), LSTM, and GRU. Comparing these models to get the best results for temperature, rainfall, and humidity prediction. The performance of the models was evaluated using standard techniques, including the Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). In individual forecasts for Maximum Temperature and Rainfall, GRU shows the lowest RMSE for Rainfall, while LSTM excels in Maximum Temperature prediction. Conv1D proves highly effective in Humidity forecasting. Our investigation emphasizes GRU’s superior performance across eight cities.

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

Deep Learning Approaches for Multi-target Weather Patterns Forecasting in Bangladesh

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sumaiya Siddiqua Mumu,
  • Syeda Samia Sultana,
  • Imranul Islam Adnan,
  • Mohammed Z. Waughfa

摘要

Weather forecasting in countries like Bangladesh poses unique challenges, given its diverse geographical features. The region experiences varying weather patterns influenced by factors such as monsoons, river systems, and coastal influences. The intricacies of high temperatures, sudden atmospheric changes, and complex circulations demand advanced forecasting methods. The research aimed to address the challenges in weather prediction for 8 cities in Bangladesh. The research used 3 deep learning models: CNN(Conv1D), LSTM, and GRU. Comparing these models to get the best results for temperature, rainfall, and humidity prediction. The performance of the models was evaluated using standard techniques, including the Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). In individual forecasts for Maximum Temperature and Rainfall, GRU shows the lowest RMSE for Rainfall, while LSTM excels in Maximum Temperature prediction. Conv1D proves highly effective in Humidity forecasting. Our investigation emphasizes GRU’s superior performance across eight cities.