A Spatio-Temporal Graph Convolutional Network for Air Quality Prediction

Round 1

Reviewer 1 Report

The paper requires extensive grammar editing.

The authors propose a novel spatio-temporal neural network, GCNInformer, that combines the graph convolution network with Informer to predict air quality data. GCNInformer incorporates information about the spatial correlations among different monitoring sites through GCN layers and acquires both short-term and long-term temporal information in air quality data through Informer layers. The GCNInformer uses MLP layers to learn low-dimensional representations from meteorological and air quality data. These designs enable GCNInformer the ability to capture the complex and nonstationary relationships between air pollutants and their surrounding environment, allowing for more accurate predictions. The experimental results demonstrate that GCNInformer outperforms other methods in predicting both short-term and long-term air quality data. The use of GCNInformer can provide useful information for air pollutant prevention and management, which can greatly improve public health by alerting individuals and communities about potential air quality hazards. 

 

Table 1 results do not appear to be statistically different for GCNInformer results compared to the other methods at any time interval. Especially given the formulations and their assumptions used.  They may be. This is an important paper and this technique is conceptually shows much promise.           

It will strengthen the results of the paper if the authors apply a statistical inference method to independently demonstrate the results of the GCNInformer are statistically different that the other baselines for predicting the PM2.5 air pollutant concentration. 

The paper requires extensive grammar editing.

Author Response

We appreciate the reviewer for giving us valuable comments, which help us to improve our manuscript. All the comments have been considered carefully, and we have revised our manuscript accordingly. The revised parts are marked up using the “Track Changes” function in the revised manuscript. The detailed point-by-point responses are listed in the attachment. The sentences highlighted in dark are the reviewers’ comments and the sentences highlighted in red are our responses.

Author Response File: Author Response.pdf

Reviewer 2 Report

In the present article entitled "A spatio-temporal graph convolutional network for air quality prediction", authors have proposed a spatio-temporal graph convolutional network (GCN Informer) and claimed that GCN Informer can capture the complex and non-stationary relationships between air pollutants and their surrounding environment, allowing for more accurate air quality predictions. To leverage the spatial and temporal correlations of air quality data, GCNInformer uses GCN layers to incorporate information about the underlying spatial correlations among different monitoring sites and Informer layers to acquire both short-term and long-term temporal information in air quality data. The performance of the proposed GCN is verified with the experimental data and found the short term and long term prediction is much accurate than other existing models. The manuscript is well written with clear description and scientifically sound enough to be accepted for publication. However, authors may consider to maintain uniform format for all equations using better equation editing tools. 

Author Response

We appreciate the reviewer for giving us valuable comments, which help us to improve our manuscript. All the comments have been considered carefully, and we have revised our manuscript accordingly. The revised parts are marked up using the “Track Changes” function in the revised manuscript. The detailed point-by-point responses are listed as follows. The sentences highlighted in dark are the reviewers’ comments and the sentences highlighted in red are our responses.

Point 1: The manuscript is well written with clear description and scientifically sound enough to be accepted for publication. However, authors may consider to maintain uniform format for all equations using better equation editing tools.

Response 1: Thanks for your valuable suggestion. We have reformatted all the equations in the manuscript using MathType software.

Round 2

Reviewer 1 Report

The authors have satisfactory addressed my comments. 

Little change since last report