Thank you very much for your insightful comments. Please find our responses as follows.

A1：Current research on prediction and early warning of urban waterlogging disaster is mainly based on physically based models, statistical methods and data-driven models. The physically based models have the advantage of interpretability, but they have strong requirements on the data of urban underlying surface and rely on complicated calculation, and thus have difficulties in large-scale applications. Statistical methods have low requirements for urban underlying surface data, but the weight settings of each factor will greatly affect the final results. The framework proposed in this paper is based on the data-driven model, which takes the advantage of the machine learning methods and the availability of huge amount of sensor-generated data. Hossein et al. presented a flood simulation framework, which used a random forest classification model and a multilayer perceptron model to identify wet or dry nodes over the domain, then estimate river depth in wet nodes. The flood simulation framework cannot predict the depth of waterlogging in real time because it is based on different relative flow grades. Our framework is based on the data-driven model, which has fitted many years of real data and has a good performance in terms of prediction accuracy and calculation efficiency.

A2：The prediction strategies used in this paper are indeed commonly used strategies in time series forecasting. However, in this paper we focus on the evaluation of these strategies with respect to the problem of urban waterlogging prediction to shed light on which strategy is more applicable and can provide more accurate results. For example, in the literature “A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition”, Ben Taieb et al. compared the performance of some common prediction strategies for the prediction of daily cash money withdrawal amounts at ATM machines. In their paper, the experiment results showed that the multi-step strategy achieved the best performance, while the Rec strategy achieved the worst performance. However, in the physical process of the change of waterlogging depth, the curve is monotonous, and therefore the Rec strategy is more adaptable.

A3：The 86.1% accuracy is the result of Rec prediction strategy. The multi-strategy here means that this framework includes the steps to select the optimal strategy, and the applicability of each prediction strategy is verified through multiple groups of experiments. The accuracy and principle of the optimal prediction strategy in the application of waterlogging prediction are explained. This is different from the concept of “coupling strategies” and does not mean that multiple strategies are combined into a new one. In future studies, we will consider coupling different strategies, modes and algorithms to improve the framework.

A4：In the literature, the accuracy of some data-driven models is below 85% in 30-minute prediction. For example, Jing H et al. proposed a novel approach to measure urban waterlogging depth by Mask R-CNN. its accuracy rate can reach 80.52%(video) and 81.38%(image). However, since the experiment setting is not the same, this conclusion may not be accurate. We will modify the manuscript to clear the confusion.

Dear Editors and Reviewers:

Thank you for your letter and for the reviewers' comments concerning our manuscript entitled "A Multi-strategy-mode-waterlogging-prediction Framework for Urban Flood Depth" (ID: nhess-2022-36). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made the correction which we hope meet with approval. The main corrections in the manuscript and the responses to the reviewer' s comments are as flowing:

Comment #1: A flowchart is necessary for the case study as well, as there have been many details regarding data extraction, preprocessing and comparison between methods, etc. in the case study.

Reply: Your comment is constructive, and we believe that a flowchart would more clearly reflect the steps of the case study and allow for a more progressive approach to the results and conclusions. We have modified the case study sections according to the process steps in the framework and expressed them in flowchart form.

Comment #2: Figure 13 to 15, if not individually discussed in detail, are suggested to be moved to appendix.

Reply: Thanks for the opinion. Given the large number of figures in this section, we have placed figures 13 to 15 in the appendix to simplify the manuscript and give a brief description.

Comment #3: There still remain grammatical errors throughout the manuscript. A thorough proofreading is needed.

Reply: Considering the reviewer's suggestion, we have revised the grammar and words throughout the manuscript to enhance the grammatical accuracy.

Based on the review comments, we rewrote the results analysis and conclusion sections of the manuscript to make the key information of this study clearer and easier to read for the reader. In the case study section, a framework flowchart step-by-step approach was used for the progressive study. The section is more clearly organized overall, from data description and processing to the application of the research methodology. To keep the reader's attention on the tasks conducted with the research objectives, parts of the conclusions and arguments poorly relevant to the research questions were removed from the manuscript. We have revised the wording and grammar of the manuscript and corrected some grammatical errors.

Thanks to the reviewers' professional comments, we could quickly target the problems and make targeted corrections. We tried our best to improve the manuscript and made some changes.

Yours sincerely,

Corresponding author: Lili Yang

E-mail: yangll@sustech.edu.cn

Dear Reviewers:

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled A Multi-strategy-mode-waterlogging-prediction Framework for Urban Flood Depth (ID: nhess-2022-36). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made the correction which we hope meet with approval. The main corrections in the manuscript and the responses to the reviewer’ s comments are as flowing:

Comment #1: First, the paper should be more clarified and concise. The introduction and literature review sections are lengthy, but fail to identify both the research gap in the current literature and the research questions to be addressed in this paper. I recommend to integrate the two sections, and clarify the research questions based on the literature review.

Reply: We have rewritten the introduction and literature review sections. On the one hand, we focus on the strengths and weaknesses of the previous literature studies, find the urgent research questions and define our research objectives in this way. On the other hand, we mainly focus on flooding risk prediction and exclude the part of the literature on risk identification and risk assessment that is not very relevant. The introduction and literature review sections are more concise and logical.

Comment #2: Second, other sections should also be shortened and presented in a more direct way. For instance, I did not get the information I expected from a conclusion section although length is enough. There are 10 tables and 15 figures, exhausting the readers to get the key information.

Reply: Sorry for the misleading content. We have carefully considered review comments, shortened the length of the data processing and result analysis sections, and removed some of the non-core bases for conclusions that may be confusing and misleading to the reader. The conclusions and the basis of the experimental results supporting the conclusions are highlighted. The results of figures 13 to 15 are briefly presented in the text, and the three figures are placed in the appendix section to make the text more concise and readable.

Comment #3: Third, I cannot understand why the authors did not include elevation data in the methodology, which should be a critical factor in determining urban floods.

Reply: Elevation data is important for flood hazard prediction and simulation because elevation directly affects surface runoff flow direction and velocity. It helps researchers to delineate catchment areas, determine watersheds and outlets, etc. The above mainly applies to flood risk prediction and simulation based on hydrodynamic methods. The manuscript is mainly devoted to solving the problem of predicting the future waterlogging depth of urban flood-prone points (determined by municipal management based on historical flooding events). It can solve the problem of the temporal distribution of urban flooding. In lines 154-156, we illustrate that the variables affecting the temporal distribution of flooding depth are mainly rainfall and previous moment flooding depth for sensor sites. The elevation data, surface type data, drainage network distribution data, etc. are constant in these flood-prone points. Hence, they can be regarded as static factors in the machine learning black box model, where the input variables are real-time rainfall data and previous waterlogging depth data, and the output variables are future waterlogging depth. Considering that the current urban ponding waterlogging sensors mainly perform limited real-time monitoring functions and lack prediction functions. Combining the historical flooding depth data of these points and implementing the model configuration, training and correction under this framework can enhance the prediction capability of future waterlogging depth at flood-prone points, which is crucial for the government to release early warning information and carry out emergency dispatch in a timely manner.

Comment #4: Finally, grammatical errors are throughout the manuscript.

Reply: Considering the reviewer’ s suggestion, we have revised the grammar and word throughout the manuscript to enhance the grammatical accuracy.

Based on the review comments, we rewrote the results analysis and conclusion sections of the manuscript to make the key messages of this study clearer and easier to read for the reader. In the case study section, a framework flowchart step-by-step approach was also used for the progressive study. The section is more clearly organized overall from data description and processing to the application of the research methodology. Some of the conclusions and their arguments that were not highly relevant to the research questions were removed from the manuscript to allow the reader to better focus on the various tasks that were conducted with the research objectives in mind. We have revised the wording and grammar of the manuscript and corrected some grammatical errors.

It was thanks to the professional comments of the reviewers that we were able to quickly target the problems and make targeted corrections. We tried our best to improve the manuscript and made some changes in it. Special thanks to you for your good comments.

Yours sincerely,

Corresponding author: Lili Yang

E-mail: yangll@sustech.edu.cn