Literature Review on Integrating Generalized Space-Time Autoregressive Integrated Moving Average (GSTARIMA) and Deep Neural Networks in Machine Learning for Climate Forecasting

Round 1

Reviewer 1 Report

In this article, the authors have reviewed the literature on generalized space-time integration of RIMA and deep neural network in machine learning for weather forecasting.

Climate change is a critical issue. The authors focus on previous work done between 2013 and 2022. The authors have developed three research questions as follows:

How does the integration of GSTARIMA-DNN model using the ML technique work?

How does the integration of GSTARIMA-DNN model utilizing ML contribute to climate data forecasting?

How does GSTARIMA-DNN model compare to GSTARIMA model in forecasting climate data?

The presentation of the article is well structured in the beginning. As a review article, the topic has significance, and the approach is well explained. Although the article is well presented and structured, I suggest the authors revise the following:

1. I recommend that the authors try to modify the structure of Table 2 so that the data provided can be better understood. For example, by adjusting the lines inside the table cells. (Table properties ... Cell... Centre).

2. 2. The quality of Figure 2 can be improved.

3. In the Results, accurately answer the three Research Questions that were mentioned in section 1.

4. In the Results, directly address the following claims that you mentioned in the introduction of the article. How have the authors proven these claims?

"The review showed that GSTARIMA-DNN integration model was a promising tool for forecasting climate in a specific region in the future.

Although Spatio Temporal and DNN approaches had been widely employed for predicting climate and its impact on human life due to their computational efficiency and ability to handle complex problems, there was currently no consensus on a universally accepted method for integrating these models that encompassed location and time dependencies.

Furthermore, it was found that GSTARIMA-DNN method, incorporating multivariate variables, locations, and multiple hidden layers, was suitable for short-term climate forecasting."

5. The authors need to confirm the validation of their assumptions and claims in the article.

Minor editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

 

This paper performed a Review on Integrating Generalized Space-Time ARIMA and Deep Neural Network in Machine Learning for Climate Forecasting, which is relatively interesting. However, I have some comments for the authors to improve the quality.

 

Major:

 

Although the authors aim to present a comprehensive in the view of the integrated approach between GSTARIMA model and DNN, most of the existing studies are not integrated GSTARIMA-DNN modeling. Current reviewed studies are either GSTARIMA or DNN relevant, rather than the hybrid method.

For example, there are no integrated GSTARIMA-DNN modeling in Fig. 1 or Table 1.

 

What are the new ideas or insights found or proposed by this work?

 

Minor:

Suggest to add some references in the introduction section.

Ham, Yoo-Geun, Jeong-Hwan Kim, and Jing-Jia Luo. "Deep learning for multi-year ENSO forecasts." Nature 573, no. 7775 (2019): 568-572.

Xu, L., Chen, N., Chen, Z., Zhang, C. and Yu, H., 2021. Spatiotemporal forecasting in earth system science: Methods, uncertainties, predictability and future directions. Earth-Science Reviews, 222, p.103828.

Qi, Y., Li, Q., Karimian, H. and Liu, D., 2019. A hybrid model for spatiotemporal forecasting of PM2. 5 based on graph convolutional neural network and long short-term memory. Science of the Total Environment, 664, pp.1-10.

Xu, L., Chen, N., Zhang, X., Chen, Z., Hu, C. and Wang, C., 2019. Improving the North American multi-model ensemble (NMME) precipitation forecasts at local areas using wavelet and machine learning. Climate dynamics, 53, pp.601-615.

Agoua, X.G., Girard, R. and Kariniotakis, G., 2017. Short-term spatio-temporal forecasting of photovoltaic power production. IEEE Transactions on Sustainable Energy, 9(2), pp.538-546.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

- Lack of clarity in research objectives: The paper lacks a clear statement of research objectives. Although the introduction section briefly mentions the purpose of compiling previous findings on Spatio Temporal and Deep Neural Networks (DNN) applied to climate data, the specific research questions are not clearly defined. The absence of well-defined research objectives makes it difficult to assess the effectiveness and contribution of the proposed integrated approach.

- Inadequate literature search methodology: The paper mentions conducting a literature search using various search engines and following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) approach. However, it does not provide details on the search strategy, inclusion and exclusion criteria, or the selection process for the included articles. The lack of transparency in the literature search methodology raises concerns about the comprehensiveness and rigour of the review.

- Lack of critical analysis and synthesis: The literature review section primarily consists of descriptive summaries of previous research without offering a critical analysis or synthesis of the findings. The authors do not provide a systematic comparison of the existing studies or highlight the gaps and limitations in the current literature. A comprehensive literature review should go beyond summarizing individual studies and should critically evaluate the existing body of knowledge.

-Insufficient discussion of the integrated approach: The paper claims to present an integrated approach between Generalized Space-Time Autoregressive Integrated Moving Average (GSTARIMA) and DNN for climate forecasting. However, the discussion on this integrated approach is limited and lacks depth. The authors do not provide sufficient information on how the integration works, the advantages it offers, or the challenges it may face. Without a detailed discussion, it is challenging to assess the novelty and effectiveness of the proposed approach.

- Lack of clarity in data analysis: The section on materials and methods briefly mentions conducting a bibliographic survey and dataset analysis. However, the paper does not provide a clear explanation of the data analysis methodology or the specific insights gained from the analysis. The results of the data analysis are presented in a visual representation without sufficient interpretation or meaningful discussion.

 

Overall, the paper lacks clarity in research objectives, methodology, critical analysis, and discussion of the integrated approach. It needs improvement before publication.

The quality of the English language in this paper could be improved. While the overall text is understandable, there are several instances of awkward phrasing, unclear sentence structures, and grammatical errors.

It is important for the authors to revise and edit the paper carefully to enhance the clarity and coherence of the language. Proofreading for grammatical and syntactical errors, as well as simplifying complex ideas and avoiding excessive technical language, would greatly improve the overall quality of the English language in this paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The paper may be accepted now.