TY  - JOUR
AU  - Ali, Ahmad
AU  - Naeem, H. M. Yasir
AU  - Ali, Riaz
AU  - Asad, Mujtaba
AU  - Heyat, Md Belal Bin
AU  - Alsarhan, Tamam
PY  - 2025
DA  - 2025/10/28
TI  - An Efficient Algorithm for Weather Forecasting Using Causal Graph Neural Network
JO  - ICCK Transactions on Advanced Computing and Systems
T2  - ICCK Transactions on Advanced Computing and Systems
JF  - ICCK Transactions on Advanced Computing and Systems
VL  - 1
IS  - 4
SP  - 258
EP  - 274
DO  - 10.62762/TACS.2025.619794
UR  - https://www.icck.org/article/abs/TACS.2025.619794
KW  - weather forecasting
KW  - causal graph learning
KW  - spatio-temporal graph neural network
KW  - attention mechanism
KW  - meteorological prediction
AB  - The rapid accumulation of large-scale, long-term meteorological data presents unprecedented opportunities for data-driven weather modeling and high-resolution atmospheric prediction. While various deep learning techniques-such as Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNNs), and Graph Neural Networks (GNNs)-have been explored for weather forecasting, the complex spatial dependencies within historical meteorological data, particularly dynamic spatial correlations among distributed observation stations, remain insufficiently addressed. To tackle this challenge, we propose a Dynamic Spatio-Temporal Fusion Graph Network (DSTFGN), a novel framework that integrates multivariate time-series analysis with graph-based causal inference to capture intricate and time-varying interdependencies among weather variables across meteorological observation networks. The DSTFGN module fuses real-time inputs (e.g., sensor data, live weather feeds, satellite observations) with historical records to model the propagation of atmospheric disturbances-such as storm systems, pressure fronts, or humidity anomalies-through the meteorological observation network. By effectively capturing dynamic spatial-temporal interactions among weather stations, our approach significantly enhances forecasting accuracy and supports adaptive meteorological monitoring strategies. Experimental evaluations on two real-world meteorological datasets demonstrate that DSTFGN consistently outperforms existing baseline models across short-term, medium-term, and long-term forecasting horizons, achieving improvements of up to 10.58\% in temperature forecasting and 8.88\% in humidity forecasting over the best competing baseline.
SN  - 3068-7969
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Ali2025An,
  author = {Ahmad Ali and H. M. Yasir Naeem and Riaz Ali and Mujtaba Asad and Md Belal Bin Heyat and Tamam Alsarhan},
  title = {An Efficient Algorithm for Weather Forecasting Using Causal Graph Neural Network},
  journal = {ICCK Transactions on Advanced Computing and Systems},
  year = {2025},
  volume = {1},
  number = {4},
  pages = {258-274},
  doi = {10.62762/TACS.2025.619794},
  url = {https://www.icck.org/article/abs/TACS.2025.619794},
  abstract = {The rapid accumulation of large-scale, long-term meteorological data presents unprecedented opportunities for data-driven weather modeling and high-resolution atmospheric prediction. While various deep learning techniques-such as Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNNs), and Graph Neural Networks (GNNs)-have been explored for weather forecasting, the complex spatial dependencies within historical meteorological data, particularly dynamic spatial correlations among distributed observation stations, remain insufficiently addressed. To tackle this challenge, we propose a Dynamic Spatio-Temporal Fusion Graph Network (DSTFGN), a novel framework that integrates multivariate time-series analysis with graph-based causal inference to capture intricate and time-varying interdependencies among weather variables across meteorological observation networks. The DSTFGN module fuses real-time inputs (e.g., sensor data, live weather feeds, satellite observations) with historical records to model the propagation of atmospheric disturbances-such as storm systems, pressure fronts, or humidity anomalies-through the meteorological observation network. By effectively capturing dynamic spatial-temporal interactions among weather stations, our approach significantly enhances forecasting accuracy and supports adaptive meteorological monitoring strategies. Experimental evaluations on two real-world meteorological datasets demonstrate that DSTFGN consistently outperforms existing baseline models across short-term, medium-term, and long-term forecasting horizons, achieving improvements of up to 10.58\\% in temperature forecasting and 8.88\\% in humidity forecasting over the best competing baseline.},
  keywords = {weather forecasting, causal graph learning, spatio-temporal graph neural network, attention mechanism, meteorological prediction},
  issn = {3068-7969},
  publisher = {Institute of Central Computation and Knowledge}
}