Scholarlink

The easy accessibility of digital meteorological data allows for more accurate forecast of complicated hydroclimatic events. However, the volume and quantity of such data, as well as the underlying complicated relationship, pose significant obstacles to conventional methods. To predict the daily maximum temperature, this study employs a hybrid deep learning (DL) technique that integrates a Convolutional Neural Network (CNN) with a Long Short-Term Memory (LSTM). The suggested technique is applied to New Delhi, an important city in India, to evaluate how well it can forecast daily maximum temperatures and predict the heatwaves. The month index along with seven meteorological parameters that are strongly related with daily temperature variance are given as an input to the suggested technique. The proposed hybrid model provides 0.07 Root Mean Square Error (RMSE), 0.94 Coefficient of Correlation (CC), and 0.88 Nash-Sutcliffe efficiency (NSE) for maximum temperature prediction.

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

Confluence of CNN and LSTM Model: A Hybrid Deep Learning Model for Heatwave Prediction

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • P. Kaviya,
  • P. Chitra

摘要

The easy accessibility of digital meteorological data allows for more accurate forecast of complicated hydroclimatic events. However, the volume and quantity of such data, as well as the underlying complicated relationship, pose significant obstacles to conventional methods. To predict the daily maximum temperature, this study employs a hybrid deep learning (DL) technique that integrates a Convolutional Neural Network (CNN) with a Long Short-Term Memory (LSTM). The suggested technique is applied to New Delhi, an important city in India, to evaluate how well it can forecast daily maximum temperatures and predict the heatwaves. The month index along with seven meteorological parameters that are strongly related with daily temperature variance are given as an input to the suggested technique. The proposed hybrid model provides 0.07 Root Mean Square Error (RMSE), 0.94 Coefficient of Correlation (CC), and 0.88 Nash-Sutcliffe efficiency (NSE) for maximum temperature prediction.