Modeling of Heavy Metal (Ni, Mn, Co, Zn, Cu, Pb, and Fe) and PAH Content in Stormwater Sediments Based on Weather and Physico-Geographical Characteristics of the Catchment-Data-Mining Approach

Round  1

Reviewer 1 Report

In general, I think the principle of this work is sound and is of importance, but I found this manuscript very difficult to read because of the way that it was written. I would recommend that the authors have it reviewed by a native English language speaker who is in the field of urban stormwater management. (As I found that some of the terminology used e.g., references to sewage connections and rainwater sediments, were not quite right from a language point of view). I would be happy to re-review this after this language edit has been made, as I think the work is novel.

Author Response

Manuscript was corrected in English language and style field

Reviewer 2 Report

The Manuscript under the title Modeling the content of heavy metals (Ni, Mn, Co, Zn, Cu, Pb, Fe) and PAHs in rain sediments based on physic-geographical characteristics of the catchment - data-mining approach was developed on the basis of research conducted on the real catchment. The article is well thought out and well prepared methodically - this is its strong side. The research subject is up-to-date and the results are presented in an interesting way.

Implementation of research in real catchment conditions is a good approach, but I have some attention here. How you determined independent parameters (concerning atmospheric conditions)? There is no explanation in the text. The lack of a clear explanation reduces the scientific value of the article.

When it comes to precipitation, there are a lot of parameters which can be adopted as independent parameters. For example intensity of rainfall (mm, mm / h), probability of occurrence, time duration, rainfall pattern, treating the catchment as a homogeneous or move of storm?

Considering that the research was carried out in real terms, one can guess that it was based on a single rain-gauge?

Nevertheless, one should pay attention to the fact that the characteristic for research in the real catchment is the lack of possibility of influencing the atmospheric parameters. Therefore, conclusions can only be drawn based on rainfall data that was "available". I consider the presented research results as research, which in the future will be extended, eg through appropriate computer simulations. In spite of the attention, I consider the research to be valuable because in my opinion it is good the theoretical basis for conducting further research towards the extension of atmospheric parameters that affect the results of pollution concentrations.

Other minor remarks:

Introduction

I  suggest to extend the introduction with rainfall parameters e.g. from Dimensioning of Required Volumes of Interconnected Detention Tanks Taking into Account the Direction and Speed of Rain Movement. In: Water  10 (12), S. 1826. DOI: 10.3390/w10121826.

Text under fig 1. Please to verify if the main channels has diameters 300-600 mm and inflows conduit 200-1000 mm? There are some underground tank or something ?

I  suggest to extend the part about PAHs for example „As consequence…. o Książek, S., Kida, M., & Koszelnik, P. (2016). The  occurrence and source of polycyclic aromatic hydrocarbons in bottom sediments of the Wisłok river. Polish J Natural Sci, 31(3),  373-386.

Methods and materials  

10th line above the table 1. “fall of catchment” ?  Probably slope of catchment?

2nd line above table 1 “catchment area is green”? Probably permeable catchment?

The sampling sites are poorly visible in the figure 1. I would suggest to  change the symbol (the dot should be bigger).

Samples collection… In this part of the Manuscript there is a lack of information about the measurement of atmospheric parameters or from where this data was collected.

In the paragraph 2.3.4.  The unit of  MAE is „%”  but in the table 3 there is some other units.

Result and discussions

Fig 4. - I would suggest to change the scale for each catchment (e.g. everywhere range from 0 - 1800) it will be easier to compare the results between particular catchment

Paragraph 3.1. Do the charts in Figure 5 contain all the results that were used to build an artificial neural network? There is also a lack of information       about number of results which was used for build of artificial neural network. What percentage of data was used for learning, validate and test. 70% 15% 15%?

When  comparison the data from charts with data presented in table 3 there is some inconsistency. Some chart look too good for such large errors (e.g.    for Co)

Author Response

General remarks

ad. 1. There are no clearly defined variables that would describe the physics of the phenomenon being studied. It has not been also determined what the impact of the catchments characteristics means (area, land use, heterogeneity of catchment) on the quality of sediments. The same applies to the dynamics of surface runoff and the characteristics of the entire stormwater drainage system (the length, capacity of the sewers and amount of storm drains in whole system) which also affect on sediments quality. An important factor in both the impact on the quality of stormwater and sediments are meteorological conditions. In many papers there is information that a significant impact on both processes is the amount of pollution accumulated on the catchment area. Therefore, the first question is – which of the processes occurring in the atmosphere are responsible for the formation of this variability, and second - which of these processes can be measured (determine their occurrence) without the need for specialized measurement systems. Determination of the dust content or concentration of pollutants is not easy and cheap at all. However, the formation of their content in the air certainly affect the phenomena that causing their removal from the air and deposition. These phenomena include rainfall, snow and the presence of fog.

ad. 2 In the chapter "Independent variables and their selection", the selection of independents variables was discussed in detail. A Fischer-Snedecor test was used for this purpose. When the calculated test probability value was smaller than p = 0.05, these variables were included in the further analyzes.

ad. 3. The main purpose of the paper was to develop a relatively simple and easy to build tool for forecasting sediment quality in stormwater drainage systems. Therefore, in presented approach we assumed that the analyzed parameters describing precipitation are homogeneous throughout the catchment (from one study site). Additionally we agree with the Reviewer that the rainfall dynamics is a crucial factor affecting the quality of stormwater and sediments. Therefore, further simulation and analysis taking into account the variability of  individual variables are planned.

ad. 4. Thank you for your valuable comments. The next step in our research will be to carry out appropriate simulations taking into account the dynamics of weather phenomena.

Other remarks

Introduction

         ad. 1. The comments of the reviewer have been included in the text of the manuscript.

ad. 2. It ought to be “1600” across the article. Text has been corrected

ad. 3. The comments of the reviewer have been included in the text of the manuscript.

Methods and materials

  ad. 1. It ought to be “slope” across the article. Text has been corrected

  ad. 2. It ought to be “permeable” across the article. Text has been corrected

  ad. 3. Figure was reedited (used symbols are bigger)

ad. 4. Text supplemented by information on measurements of atmospheric parameters

  ad. 5. All units have been standardized

Result and discussion

ad. 1. Changing the scale of figures will cause some of them to be unreadable (only the points filling the small part of the figure will be visible). It was therefore decided on other scales.

ad. 2. The information concerning percentage data used for learning, validation and test was added. 

ad. 3. Graphs are correct. The high relative error is due to the comparison of the relatively small observed and predicted value of the examined indicators (for example for Co: predicted value was 0.1 mg/kg, while observed value was close to 0 mg/kg – in that case the error is very large).

Reviewer 3 Report

Authors mention the model and can author link the data with a numerical watershed-water quality model, such as Xia et al., 2011?

Xia, M., Craig, P.M., Wallen, C.M., Stoddard, A., Mandrup-Poulsen, J., Peng, M., Schaeffer, B., Liu, Z. (2011). Numerical simulation of salinity and dissolved oxygen at Perdido Bay and adjacent coastal ocean. Journal of Coastal Research. 27(1), 73-86.  doi: http://www.jcronline.org/doi/abs/10.2112/JCOASTRES-D-09-00044.1  

What is the board application of author's method? Can it be applied to other areas? Overall, I agree it is a nice paper and could be published after minor revision. 

Author Response

The comments of the reviewer have been included in the text of the manuscript.

Round  2

Reviewer 1 Report

I found the manuscript to be much improved. 

I think this is a novel study that should be published for the following reasons:

 - the methodology presented can be extrapolated to other situations and as such provides an innovative way to deal with the problem of spatio-temporal variability in sediment quality

 - it provides some findings about the key determinants of sediment quality, which are not unexpected, but can then be used to prioritise management actions in urban environments - not just in the case study city, but in other cities.