the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
From gridded runoff to streamflow: application of statistical post-processing to generate seasonal streamflow forecasts
Abstract. Hydrological models can be categorized as either fully distributed grid-based models or catchment-based models coincident with river gauging stations. Grid-based models provide national coverage at a relatively high spatial resolution. Catchment-based models target specific catchments upstream of a river-gauging station. While not providing seamless national coverage, catchment-specific model calibration allows catchment-based models to achieve improved performance relative to a distributed model using a single optimized parameter set. Catchment-based models often rely on the supply of real-time hydrological observations, the latency of which hinders the efficiency and timelessness of forecast generation and publication. This study evaluates whether a nation-wide grid-based hydrological model, coupled with statistical post-processing, generates comparable forecast skill to that of a catchment-based statistical hydrological model.
Two hydrological models are evaluated at 449 gauging stations across Australia. The Australian Water Resource and Assessment model (AWRA-L) is a 5km2 distributed landscape model that provides seasonal forecasts of key hydrological variables (soil-moisture and runoff) at the monthly temporal resolution. The necessary climate forcing is provided by the ACCESS-S seasonal climate forecasting model. Coupled with forecast post-processing, seasonal forecasts of river discharge (streamflow) can be generated from AWRA-L seasonal forecasts at gauged catchments across Australia. The second model evaluated in this study is a statistical hydrological model that is calibrated locally at the same individual catchments, where observed catchment conditions (antecedent observed streamflow) is the primary input variable with which to forecast streamflow.
Results from this study indicate that seasonal forecasts from a distributed hydrological model coupled with statistical post-processing achieve similar forecast skill to a locally calibrated statistical hydrological model where the observed catchment conditions represented by observed antecedent streamflow is the only input variable. The statistical post-processor is calibrated against historical streamflow observations but excludes antecedent observed streamflow as an input variable. For many catchments, AWRA-L can represent initial catchment conditions as well as the hydrological response arising from climate forcing from a seasonal climate model. The post-processing model provides a bias-correction that is a function of the magnitude of a hydrological model response only (e.g. runoff).
Forecast performance of the AWRA-L model is highest in the high-flow season across Australia but is limited in the low-flow season in certain regions (e.g. the tropical dry season in northern Australia). Inclusion of antecedent streamflow observations as an input variable results in excellent performance due to strong hydrological persistence in low flows. Depending on the location and time of year, either root-zone soil-moisture or runoff are found to maximise forecast skill. Root-zone soil-moisture maximises forecast skill for a higher proportion of sites than runoff overall, particularly in the low-flow season. As forecast skill increases, the proportion of sites for which runoff maximises forecast skill increases, particularly in the high-flow season. Results also indicate the degree to which climate forcing from a seasonal climate model contributes to forecast skill, as distinct to initial catchment conditions and hydrological persistence.
- Preprint
(3654 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-1379', Anonymous Referee #1, 04 Jun 2025
-
AC1: 'AC reply on RC1', Christopher Pickett-Heaps, 26 Sep 2025
The authors would like to thank the reviewer #1 for their suggestions and helpful feedback. We will be happy to improve the manuscript in the following areas:
- Shorten the abstract
- Redraft the introduction to more clearly articulate the objectives, scope and innovations of the study
- More clearly describe the purpose in which the BJP model is used, which are:
- BJP as a stand-alone hydrological forecasting model
- BJP as a statistical forecast post-processor
- Reduce the length of the manuscript as much as possible and remove all appendices.
We will address the reviewer’s comments as follows.
Forecast post-processing plays an important part in improving the skill of raw forecasts generated by numerical models. In this paper, the attention is paid to post-processing raw streamflow forecasts generated by the Australian Water Resource and Assessment model (AWRA-L). In general, the manuscript has plenty of methods and results. On the other hand, it seems that the methods and results are not clearly presented.
Reading through the manuscript, it is found that the manuscript is more like a report instead of a journal paper. There are six comments for further improvements of the paper:
We appreciate the fact that this is a long paper with multiple components described in the methods section. The authors would be happy to shorten the paper as much as possible, including removal of the appendices. These were included as a convenience to the reader. However, to shorten the paper, readers can access the cited publications.
First of all, while hydrological and statistical models are two types of models, it seems that the authors have mixed these two. In the second paragraph, it is stated that “Two hydrological models are evaluated … The Australian Water Resource and Assessment model (AWRA-L) is … The second model evaluated in this study is a statistical hydrological model …” It is pointed out that the “statistical hydrological model” cannot be simply considered to be a hydrological model.
The terminology used in the manuscript reflects existing literature (e.g. Zhao et al. 2015 and references therein). It is appropriate to use the term statistical hydrological model to help distinguish it from a process-based hydrological model (e.g. AWRA-L, GR4J) while emphasising the fact that it is being used for the same purpose – hydrological forecasting.
The distinction (label) of a BJP model configured for hydrological forecasting (a statistical hydrological model) is important so not to confuse the same BJP statistical model configured as a statistical hydrological post-processing model used to correct AWRA-L. We believe that the labels ‘statistical hydrological model’ and ‘statistical post-processor' is thus appropriate.
The second comment relates to the first one. If the “statistical hydrological model” is an established model, the authors may want to pay attention to the post-processing of the raw AWRA-L forecasts. That is, the methods and results should clearly illustrate how raw forecasts are generated by the AWRA-L model (a dynamical model) and then post-processed by statistical models.
The application of BJP depends upon how the statistical model is configured. In this study two primary configurations are used:
(1) Hydrological statistical model:
- Input predictor variable: Antecedent observed streamflow
- Output predictand variable: Forecast streamflow
(2) Hydrological statistical post-processor
- Input predictor variable: Raw hydrological model forecasts (e.g. runoff)
- Output predictand variable: Forecast streamflow
The two configurations are important for this study. Configuration (1) is indeed an established model but it is also a benchmark model to which we compare to the AWRA-L forecasts with forecast post-processing. It is also a ‘locally calibrated’ model using real-time local observations to forecast streamflow. However, the focus of the paper is to evaluate BJP as a forecast post-processing model (configuration 2).
In response to the reviewer comment:
“...the methods and results should clearly illustrate how raw forecasts are generated by the AWRA-L model and then post-processed by the statistical models”:
Describing in detail how raw forecasts are generated by AWRA-L lies outside the scope of this study. The description of AWRA-L has been published previously, particularly the AWRA-L landscape model itself (Frost et al. 2018) and the generation and verification of seasonal hydrological models (Vogel et al. 2021, Pickett-Heaps & Vogel 2022). We believe we adequately describe the application of forecast post-processing to the AWRA-L raw forecasts, the primary focus of the paper.Thirdly, the flowchart in Figure 1 fails to present how the Bayesian Joint Probability (BJP) model is implemented in post-processing. There are two BJP models – “BJP model calibration” and “calibrated BJP statistical model”. It is noted that the BJP model is well-known in the field of statistical hydrology. Given that it is applied to calibrate raw forecasts, why is there a “calibrated BJP statistical model”? In the meantime, it is pointed out that there is no mathematical formulation of the BJP model. Given that it is a statistical model, mathematical formulation of marginal and joint distributions are necessary.
The flowchart in figure 1 depicts the same BJP model under the same configuration as a forecast post-processor. The figure does not depict two separate BJP models. The BJP model must first be calibrated before it can be used to post-process a forecast from AWRA-L. As stated in the manuscript, the BJP post-processing model is calibrated with data from two datasets:
- Monthly hindcasts (retrospective forecasts) from a hindcast dataset spanning 37 years (1981-2017)
- Monthly observed streamflow (aggregated from daily/sub-daily hydrological observations)
Once the BJP post-processor is calibrated, it can be applied to a real-time AWO forecast. If it is applied to a specific hindcast (for forecast verification purposes), the calibration dataset must exclude this specific hindcast (and hindcasts from as 5-year period from the date of the specific hindcast). A separate BJP model calibration is required for every hindcast in the AWO hindcast dataset to generate a robust verification dataset with which to assess forecast skill.
We are happy to improve the labelling in figure 1 and add text in the figure caption to avoid confusion. The important point here is the BJP model depicted in figure 1 is a BJP model using the same configuration as required for forecast post-processing only.
The reviewer states that the mathematical formulation of the marginal and joint distributions are necessary. We are unsure what we have neglected to include. Furthermore, a formal and detailed description of the BJP model as used in hydrological modelling not a focus of this paper and not an original development. Relevant papers include Zhao et al. (2015), Wang et al. (2009, 2011).
Fourthly, it is confusing that both climate and hydrological forecasts are illustrated in the methods, in particular in Section 2.1. If the attention is on hydrological forecasts, then climate forecasts that are used to drive hydrological models should never be highlighted. For the
We respectfully disagree and would appreciate a reason(s) for why climate forecasts should never be highlighted. The climate model, providing the climate input forcing to the hydrological model, is a key component in the process of generating seasonal forecasts of streamflow that are influenced both by:
- Evolving catchment conditions (represented by the model states within AWRA-L)
- Climate input forcing
The discussion of the climate model is brief – only 1 paragraph. More detail is included in relevant citations. The climate model is important because the uncertainty in the AWRA-L hydrological forecast ensembles is driven by the uncertainty in the climate forecast ensembles.
Fifthly, the figures in this paper are not of the same format. The authors may want to generate the figures in the same software or using the same programming language.
All figures use the same programming language (Python) and libraries (specifically Matplotlib) to generate all the figures in this manuscript. We are unsure in what ways the figures are not of the same format.
Sixthly, the authors are suggested to create a supplementary material. The lengthy descriptions on the methods can be put to the supplementary material. In this paper, the focus can be concentrated on “new” things, instead of existing methods. In the meantime, the appendices can also be put to the supplementary material.
We are happy to consider redrafting this manuscript to shorten it as much as possible for readability while maintaining all major threads of the manuscript.
The appendices were included primarily as a convenience to the reader. However, much of the material can be accessed from relevant citations in this paper. The reader can access this additional information from these published papers/reports rather than accessing supplementary material. We therefore propose to remove the appendices from the manuscript.
Below are a few minor comments:
- “5km2” on Line 21 is informal. “km2” does not represent square kilometer.
Noted and will change
- There is usually one paragraph for the abstract. Currently, the abstract has 4 paragraphs. It is too lengthy to highlight the methods/findings.
We will shorten the abstract.
- The legend of Figure 4 on Page 9 is too small to be seen.
We will increase the size of the legend.
- Section 2.5 has only one sentence – “This section describes the input datasets required for the BJP model calibration described in section 2.4.” Are you serious when writing the paper?
We apologise for this. The subsequent two sections were in fact sub-sections of section 2.4 which should have been clear given the headings of the subsequent sections. This was caused by a glitch in the word processing software.
- The tables in this paper are of different format. For example, there is no head in Table 6.
We can work to improve consistency in format of all tables in the manuscript.
- The bullet points in this paper are usually for POWERPOINT and rarely seen in WORD. For example, there are 7 bullet points on Page 13.
These bullet points are generated by MS Word, not PowerPoint. No specification of bullet-point format is given by HESS for submitted manuscripts (as opposed to final drafting).
- On Page 26, the section of Conclusion consists of 6 bullet points. It is not serious.
We can rewrite the conclusions as a paragraph.
Citation: https://doi.org/10.5194/egusphere-2025-1379-AC1
-
AC1: 'AC reply on RC1', Christopher Pickett-Heaps, 26 Sep 2025
-
RC2: 'Comment on egusphere-2025-1379', Anonymous Referee #2, 17 Aug 2025
2025.8.17 Review for HESS (EGUsphere-2025 -1379)
1. Overall comments
This study reports seasonal foresting of streamflow from a national operation system. There is a strong reporting sense, not like a research paper, and I am struggling to understand the overall research objectives, and scientific findings. The research gaps are not well defined. I currently can not meet high standards required by HESS journal
2 . Abstract: It is too long to be suitable for a scientific journal like HESS. (1- 2 sentences for backgrounds, 1 sentence for knowledge gaps, 1-2 for methods, 3-4 sentences for results, and 1-2 implications)
3. Introduction
- It is localized introduction. More generalized introduction is required for international audiences
- Why should we put physically hydrological modelling results outputs into a statistical modelling approach (BJP) for seasonal streamflow forecasting? I really do not understand the logic. The physically hydrological models are run at each 5-km resolution across the continental scale, which can be easily used for forecasting in the real-time and seasonal scale. It is more important for using the physically based modelling results in ungauged basins (or catchments). While BJP model is widely used for seasonal forecasting at gauged catchments with inputs that are independent of the streamflow (i.e. atmospheric circulation index, catchment attributes, etc)?
- Please articulate the innovation points
- Pease articulate the specific research knowledge gaps methods
- Methods
- It is very confusing to introduce BJP model and its calibration processing. Why should BJP model require AWRL-L outputs since it is calibrated against streamflow and it can only be applied to the local catchment for streamflow forecasting.
- For a fair comparison, I suggest an independent approach use the BJP model that takes inputs that are nothing to do with AWRA-L model, Then, the BJP model forecasting results can be compared with AWRA-L forecasting results?
- Testing in ungauged basins. BJP model has not be applied in ungauged catchment, which makes the scientific merits degraded.
I strongly suggest that the authors rephrase the paper to fit for a high-standard international journal. There is too tedious local introduction which is not required internationally. However, there are lack of common elements (concise writing, strengthening innovation and knowledge gaps, deep discussion, and concise conclusions) required by high standards of international journals.
Citation: https://doi.org/10.5194/egusphere-2025-1379-RC2 -
AC2: 'AC reply on RC2', Christopher Pickett-Heaps, 26 Sep 2025
The authors would like to thank the reviewer #2 for their suggestions and helpful feedback. We will be happy to improve the manuscript in the following areas:
- Shorten the abstract
- Redraft the introduction to more clearly articulate the objectives, scope and innovations of the study
- More clearly describe the purpose in which the BJP model is used, which are:
(1) BJP as a stand-alone hydrological forecasting model
(2) BJP as a statistical forecast post-processor
- Reduce the length of the manuscript as much as possible and remove all appendices.
We will address the reviewer’s comments as follows.
1. Overall comments
This study reports seasonal foresting of streamflow from a national operation system. There is a strong reporting sense, not like a research paper, and I am struggling to understand the overall research objectives, and scientific findings. The research gaps are not well defined. I currently can not meet high standards required by HESS journal.
We will redraft the introduction to more clearly articulate the objectives of this manuscript and to reduce the ‘localised’ introduction.
2 . Abstract: It is too long to be suitable for a scientific journal like HESS. (1- 2 sentences for backgrounds, 1 sentence for knowledge gaps, 1-2 for methods, 3-4 sentences for results, and 1-2 implications).
We acknowledge the abstract is lengthy, we will edit the abstract to reduce the length.
3. Introduction
It is localized introduction. More generalized introduction is required for international audiences
Why should we put physically hydrological modelling results outputs into a statistical modelling approach (BJP) for seasonal streamflow forecasting? I really do not understand the logic. The physically hydrological models are run at each 5-km resolution across the continental scale, which can be easily used for forecasting in the real-time and seasonal scale. It is more important for using the physically based modelling results in ungauged basins (or catchments). While BJP model is widely used for seasonal forecasting at gauged catchments with inputs that are independent of the streamflow (i.e. atmospheric circulation index, catchment attributes, etc)?
Please articulate the innovation points
Pease articulate the specific research knowledge gaps methods
We are happy to redraft the introduction to not be too ‘localised’. However, we respectfully disagree that international audiences would not be interested in developments undertaken by one of the largest National Meteorological and Hydrological Service agencies, in collaboration with Australian government research organisations and the university sector.
It is an incorrect assertion that a physically-based hydrological model run at 5km spatial resolution should be sufficient to model river discharge at the sub-seasonal to seasonal scale. Likewise, it is incorrect to assume that spatial resolution alone determines the skill/accuracy of a hydrology model. We note that the other published review of this manuscript opens with:
"Forecast post-processing plays an important part in improving the skill of raw forecasts generated by numerical models"We have results that show raw model output from AWRA-L (e.g. runoff calculated for the upstream catchment of a gauging stations) is significantly biased when compared to gauged streamflow at the same gauging station. Due to the length of the paper, we have not specifically included these results, but this can be seen in Frost et al., 2018. Even locally calibrated hydrological models require post-processing either in simulation mode or forecast mode (e.g. Pokhrel et. al 2012, McInerney et al. 2020, Woldemeskel et al. 2018).
There are multiple reasons for why raw model output from AWRA-L (specifically ‘runoff’) requires forecast post-processing:
- Continental model calibration using streamflow. AWRA-L is calibrated/parameterised using a range of data (including streamflow from ~300 gauging stations) to generate single parameter dataset that is used to model the hydrological surface water balance across the continent. It is therefore no surprise that raw AWRA-L output (runoff) is biased and requires a correction.
- The input forcing from a climate forecasting model contains inherent inaccuracies that lead to errors in runoff when compared to observed river discharge.
- Inherent inaccuracies in the AWRA-L itself (e.g. parameterisations of hydrological processes not directly modelled) lead to inaccuracies. Thus, even when AWRA-L is forced with 'observed' climate grids ('perfect' climate input forcing), model errors will arise that must be corrected for through a post-processing model.
Many studies demonstrate the need for both forecast post-processing and hydrological simulation post-processing (with observed input forcing). This is even the case with locally calibrated models such as GR4J (so-called rainfall-runoff models). Many post-processing models use observed antecedent conditions to apply a model correction. In this manuscript, we demonstrate that it is possible to achieve acceptable forecast skill using a post-processing model that is not dependant on observed antecedent conditions, which in turn requires a real-time hydrological observation data feed. This greatly improves the flexibility and reliability of a forecasting system.
We would be happy to update the introduction of the manuscript to more clearly articulate the objectives of this study.
The reviewer states that:
"While BJP model is widely used for seasonal forecasting at gauged catchments with inputs that are independent of the streamflow (i.e. atmospheric circulation index, catchment attributes, etc)?"It is unclear if the reviewer is making a statement or asking a question. In response:
The BJP model is used for seasonal hydrological forecasting in Australia. However, the primary input predictor variable is streamflow (specifically, antecedent observed streamflow). While other studies may use BJP with inputs independent of streamflow, it is virtually guaranteed that forecast skill will be improved by incorporating streamflow as an input predictor variable. Respectfully, the reviewer could at least provide some references of such studies.
We completely agree with the reviewer that forecasting at ungauged catchments is an admirable long-term objective. However, we must first demonstrate the skill of the model at gauged stations. This manuscript, as well as previously cited studies, clearly demonstrates that the errors apparent in seasonal forecasts from AWRA-L when comparing to gauged streamflow necessitates forecast post-processing.
The reviewer also neglects to comment on a major finding of the study that root-zone soil-moisture in fact provides greater forecast skill relative to runoff. This is the case both in ‘dryer’ catchments and within the ‘lower’ flow regimes (i.e. low streamflows in the dry months of the year) of otherwise ‘wet’ catchments.
- Methods
It is very confusing to introduce BJP model and its calibration processing. Why should BJP model require AWRL-L outputs since it is calibrated against streamflow and it can only be applied to the local catchment for streamflow forecasting.
For a fair comparison, I suggest an independent approach use the BJP model that takes inputs that are nothing to do with AWRA-L model, Then, the BJP model forecasting results can be compared with AWRA-L forecasting results?
Testing in ungauged basins. BJP model has not be applied in ungauged catchment, which makes the scientific merits degraded.
In response to the claim 'Why should BJP model require AWRA-L outputs':
The AWRA-L hydrological model is the primary, physically-based model with which to generate a hydrological forecast, with input forcing from a seasonal climate model. However, AWRA-L raw output (runoff) is significantly biased. Consequently, we require a forecast post-processing model (e.g. BJP) to provide a correction such that accurate (relatively un-biased) estimates of forecast streamflow (river discharge) can be obtained. To achieve this, BJP is calibrated against a historical streamflow observations.
Note here that BJP does not include real-time observations of streamflow (observed initial ‘catchment conditions’) as an input variable. That means a correction is applied based on runoff magnitude only. Runoff itself is influenced by the initial states (catchment conditions) of the AWRA-L model itself and not from observed initial states provided by a real-time hydrological observation data feed. This is an important finding, particularly given AWRA-L is not parameterised locally for individual catchments (a single optimised parameter dataset is used across Australia).
The reviewer states:
'For a fair comparison, I suggest an independent approach use the BJP model that takes input that are nothing to do with AWRA-L model.
This is exactly what we have done, as described in section 2.9, table 7. The benchmark model (based on BJP using antecedent observed streamflow to forecast streamflow) is both (1) completely independent of AWRA-L and (2) locally calibrated, using observed antecedent catchment conditions (recently observed streamflow) to forecast streamflow.
The reviewer states:
'Testing in ungauged basins. BJP model has not been applied in ungauged catchment, which makes the scientific merits degraded.'
In response:
- 'Testing' at ungauged catchments is outside the scope of the study. We believe that before any attempt to do this is initiated, knowledge of how well the AWRA-L model performs at gauged stations is necessary. We find that forecast post-processing is required to either match or exceed the forecast skill from the benchmark model (independent of AWRA-L). This result alone points to the scientific merits of the manuscript.
- The reviewer doesn’t elaborate on how 'testing' at ungauged stations might be achieved, including relevant citations of a potential methodology that could be applied using a model such as BJP. Similarly, it is unclear what 'data' would be used to both calibrate and test the BJP model at ungauged catchments.
- In relation to ungauged catchments, a key question instead is how to make use of the results presented in this manuscript to generate forecasts at ungauged catchments and how best to characterise the uncertainty of these forecasts at ungauged catchments.
I strongly suggest that the authors rephrase the paper to fit for a high-standard international journal. There is too tedious local introduction which is not required internationally. However, there are lack of common elements (concise writing, strengthening innovation and knowledge gaps, deep discussion, and concise conclusions) required by high standards of international journals.
Given the confusion over the aims and scope of the study, we will rephrase/rewrite the introduction so that the objectives of the study and innovations achieved are more clearly articulated. We will redraft the manuscript in relation to improvements suggested by the reviewer (e.g. concise writing). While the reviewer notes a lack of ‘deep discussion’, the reviewer has not addressed specifically what is both missing and redundant from the discussion in the current manuscript. We will nonetheless endeavour to improve the discussion. Similarly, the reviewer does not address why the current set of conclusions, provided as a series of bullet-points for increased clarity, do not meet the required standards of an international journal.
Citation: https://doi.org/10.5194/egusphere-2025-1379-AC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
682 | 58 | 19 | 759 | 12 | 24 |
- HTML: 682
- PDF: 58
- XML: 19
- Total: 759
- BibTeX: 12
- EndNote: 24
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
Forecast post-processing plays an important part in improving the skill of raw forecasts generated by numerical models. In this paper, the attention is paid to post-processing raw streamflow forecasts generated by the Australian Water Resource and Assessment model (AWRA-L). In general, the manuscript has plenty of methods and results. On the other hand, it seems that the methods and results are not clearly presented.
Reading through the manuscript, it is found that the manuscript is more like a report instead of a journal paper. There are six comments for further improvements of the paper:
First of all, while hydrological and statistical models are two types of models, it seems that the authors have mixed these two. In the second paragraph, it is stated that “Two hydrological models are evaluated … The Australian Water Resource and Assessment model (AWRA-L) is … The second model evaluated in this study is a statistical hydrological model …” It is pointed out that the “statistical hydrological model” cannot be simply considered to be a hydrological model.
The second comment relates to the first one. If the “statistical hydrological model” is an established model, the authors may want to pay attention to the post-processing of the raw AWRA-L forecasts. That is, the methods and results should clearly illustrate how raw forecasts are generated by the AWRA-L model (a dynamical model) and then post-processed by statistical models.
Thirdly, the flowchart in Figure 1 fails to present how the Bayesian Joint Probability (BJP) model is implemented in post-processing. There are two BJP models – “BJP model calibration” and “calibrated BJP statistical model”. It is noted that the BJP model is well-known in the field of statistical hydrology. Given that it is applied to calibrate raw forecasts, why is there a “calibrated BJP statistical model”? In the meantime, it is pointed out that there is no mathematical formulation of the BJP model. Given that it is a statistical model, mathematical formulation of marginal and joint distributions are necessary.
Fourthly, it is confusing that both climate and hydrological forecasts are illustrated in the methods, in particular in Section 2.1. If the attention is on hydrological forecasts, then climate forecasts that are used to drive hydrological models should never be highlighted. For the
Fifthly, the figures in this paper are not of the same format. The authors may want to generate the figures in the same software or using the same programming language.
Sixthly, the authors are suggested to create a supplementary material. The lengthy descriptions on the methods can be put to the supplementary material. In this paper, the focus can be concentrated on “new” things, instead of existing methods. In the meantime, the appendices can also be put to the supplementary material.
Below are a few minor comments: