Fine-grained Conditional Convolution Network with Geographic Features for Temperature Prediction

• Published in: IEEE transactions on geoscience and remote sensing Vol. 61; p. 1
• Main Authors: Zhang, Cheng, Zhao, Guoshuai, Liu, Junjiao, Zhang, Xingjun, Qian, Xueming
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Conditional convolution; Convolution; Data models; Deep neural network; Geography; Kernel; Machine learning; Mathematics; Meteorological data; Meteorology; Multi-scale feature; Physics; Predictions; Predictive models; Task analysis; Temperature; Temperature distribution; Temperature prediction; Weight
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2023.3298318

Short-to-medium term temperature prediction in high resolution is a very challenging task, involving meteorology, physics, mathematics, geography, and many other subjects. Its purpose is to fit a complex function from historical meteorological data to predict the future 1-5 days temperature, which is a typical spatio-temporal prediction problem. Meteorological data show complex correlations in local space. Most of the existing machine learning methods are based on image pixel-level tasks or spatio-temporal prediction tasks, which model meteorological data without considering the characteristics of meteorological data and use rough global patterns to model local space which would lose many details. To address the above issues, our work fine-grained conditional convolution network (FCCN) proposes a novel grid-level conditional convolution module, including a local geographic adaptive weight and a local data adaptive weight. These two components are integrated into a multi-scale meteorological fusion GRU architecture for the end to end temperature prediction. Experiments in real-world datasets from ERA-5 show our FCCN model has a better performance than all other baseline methods.