Intra-city Scale Graph Neural Networks Enhance Short-term Air Temperature Forecasting

Wang, Han; Tang, Jianheng; Zhang, Jize; Yang, Jiachuan

doi:https://doi.org/10.5194/egusphere-2025-3429

Preprints

https://doi.org/10.5194/egusphere-2025-3429

Preprints

25 Aug 2025

| 25 Aug 2025

Intra-city Scale Graph Neural Networks Enhance Short-term Air Temperature Forecasting

Han Wang, Jianheng Tang, Jize Zhang, and Jiachuan Yang

Abstract. Air temperature (T_a) has critical implications for various socioeconomic sectors, yet its dynamics are particularly complex in urban areas due to heterogeneous built environments, landscapes, and diverse anthropogenic activities. Physics-based models struggle with intra-city T_a forecasts due to inadequate urban representation and limited spatial resolution. While weather observation networks offer promising alternatives for direct local T_a modeling, an effective framework to leverage these intra-city data remains lacking. Here, we demonstrate that graph neural networks (GNNs) can harness observation network information to refine T_a prediction at individual locations and elucidate underlying mechanisms. Our novel Mix-n-Scale framework with GNNs achieves over 12 % improvement in short-term T_a forecasts compared to conventional time-series approaches. Further model evaluation disentangles performance variations with local T_a variability in diverse spatiotemporal contexts, indicating distinct patterns of intra-city heterogeneity across seasonal and diurnal scales. Our findings establish graph-based approaches for leveraging proliferating urban sensor data and advancing understanding of T_a spatiotemporal dynamics in complex urban environments.

Received: 16 Jul 2025 – Discussion started: 25 Aug 2025

Competing interests: The authors declare no conflicts of interest relevant to this study.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 3078 KB)

Supplement (1029 KB)

Download & links

Han Wang, Jianheng Tang, Jize Zhang, and Jiachuan Yang

Status: final response (author comments only)

RC1:
'Comment on egusphere-2025-3429', Anonymous Referee #2, 16 Sep 2025
This study introduces a novel Mix-n-Scale framework with GNNs for short-term air temperature (Ta) forecasting at the intra-city scale. The authors convincingly demonstrate that incorporating spatial interactions improves forecast skill, achieving more than a 12 % reduction in RMSE compared with a conventional LSTM-only baseline. They provide a thorough evaluation of model performance and present well-structured analyses. Overall, this is a solid paper and is a meaningful contribution with clear potential to advance short-term forecasting in urban areas. However, there are several areas where additional experiment and discussion would strengthen the manuscript. I would support publication after minor revisions.
Major comments:
The authors show that adding GNN-based spatial aggregation improves performance, but it is unclear how much of the gain depends on the number, proximity, and connectivity of stations. I suggest including the following analyses to provide deeper insight into the value of spatial information for Ta forecasting and to offer practical guidance for applying the framework in settings with sparser or differently configured station networks.
Conduct a sensitivity analysis by removing stations one by one and evaluate how forecast performance changes. For example, I would like to know the effectiveness of inclusion of station 2 on station 3 given their distance. This would also help justify the assumption that “clear directed relationships for information propagation from specific ‘super-nodes’ may not exist” (Lines 236–237).

Verify whether the observed performance gains arise from the GNN’s relational structure rather than simply having access to neighbor data. Consider comparing with a simpler baseline where neighbor embeddings are concatenated or averaged and fed directly to the final prediction layer without using GNN.

Discuss why including global predictors appears to worsen GSAGE performance (Fig. S1) and explain the rationale for keeping them.

Since the manuscript discusses drawbacks of reanalysis-trained DL models (Lines 49 – 51), please add a brief benchmarking context (e.g., a small literature table or paragraph) showing how your error metrics compare with these approaches.

The ensemble approach is a key novelty and clearly improves performance over any single model. It would be helpful to know how sensitive the results are to the ensemble size or to variations in the hyperparameters, as well as how computational resources scale with ensemble complexity to inform an optimized training strategy. What are the best-performing topologies of the GNN and how different are they from each other?

The explanation of the spatiotemporal pattern of forecast performance in Section 3.4 could be clearer. While the authors report RMSEs, local Ta variability, and site characteristics, they do not clearly connect these characteristics to the observed performance differences. For instance, the statement that “the pronounced diurnal contrast can be primarily attributed to solar radiation-induced perturbations and consequent atmosphere-land interactions, highlighting the inherent challenges in capturing daily peak values” (Lines 323 – 325) notes the difficulty but does not explain why these processes make the peak harder to forecast. I recommend revising this section to make those linkages more explicit.

Minor comments:
Latitude and longitude swapped in Table 1.

Since GAT and GSAGE themselves are GNNs without LSTM encoding, the labels or figure captions could be clarified to avoid misunderstanding. For example, Fig. 1 may read as though you are comparing a standalone GNN with an LSTM, rather than GNNs applied on top of LSTM embeddings.

Line 211: revise “the mean Ta patterns is …” to “the mean Ta patterns are …”.

Line 266: revise “significantly outperform than …” to “significantly outperform the …”.
Citation: https://doi.org/10.5194/egusphere-2025-3429-RC1
RC2: 'Comment on egusphere-2025-3429', Yuanjian Yang, 04 Oct 2025

This work demonstrated that graph neural networks (GNNs) can harnessobservation network information to refine Ta prediction at individual locations and elucidate underlying mechanisms. The topic is very interesting and has important implications in prediction of urban weather. The paper is well organized and written. The findings of this study are worth of publication in the ACP after minor revision as following:
1. The rolesof Urban morphology, local circulation, anthropogenic heat release and land use and their impacts on local climate in the Hong Kong should be compared with other regions, that is, can LOI defined in the present work be related to these local background factors? Just for wind or altitude?

Effects of anthropogenic heat release upon the urban climate in a Japanese megacity; Diurnal variation of amplified canopy urban heat island in Beijing megacity during heat wave periods: Roles of mountain-valley circulation and urban morphology; Simulating the Regional Impacts of Urbanization and Anthropogenic Heat Release on Climate across China;Unevenly spatiotemporal distribution of urban excess warming in coastal Shanghai megacity, China: Roles of geophysical environment, ventilation and sea breezes;Adjustment of the urbanization bias in surface air temperature series based on urban spatial morphologies and using machine learning; Impacts of Various Local Climate Zones on Canopy Urban Heat Island and Dry Island: Spatial Heterogeneity and Relative Contributions in Beijing
2. About overfitting issues of machine learninglearning models, how did you address or aovid it?
3. Station ID can be removed in Figures 2 and 6b for clear show.

Citation: https://doi.org/10.5194/egusphere-2025-3429-RC2

Han Wang, Jianheng Tang, Jize Zhang, and Jiachuan Yang

Supplement

https://doi.org/10.5194/egusphere-2025-3429-supplement

Han Wang, Jianheng Tang, Jize Zhang, and Jiachuan Yang

Viewed

Total article views: 1,807 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
1,766	34	7	1,807	20	18	21

HTML: 1,766
PDF: 34
XML: 7
Total: 1,807
Supplement: 20
BibTeX: 18
EndNote: 21

Views and downloads (calculated since 25 Aug 2025)

Month	HTML	PDF	XML	Total
Aug 2025	782	8	0	790
Sep 2025	948	12	4	964
Oct 2025	36	14	3	53

Cumulative views and downloads (calculated since 25 Aug 2025)

Month	HTML	PDF	XML	Total
Aug 2025	782	8	0	790
Sep 2025	948	12	4	964
Oct 2025	36	14	3	53

Viewed (geographical distribution)

Total article views: 1,803 (including HTML, PDF, and XML) Thereof 1,803 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 09 Oct 2025

Short summary

Air temperature (T_a) varies substantially within cities, yet physics-based models often struggle to capture this fine-scale variability. We demonstrate that a deep learning graph-based approach—leveraging observational sensor networks—can significantly enhance short-term T_a forecasting by refining local amplitude patterns. Furthermore, we introduce an automated framework to streamline graph construction. This study also demonstrates the distinct spatiotemporal dynamics of intra-city T_apatterns.


Total:	0
HTML:	0
PDF:	0
XML:	0