Assessment of an Alternative Climate Product for Hydrological Modeling: A Case Study of the Danjiang River Basin, China
, Wenfeng Ding 1,2,*, Wentao Xu 1,2, Xiudi Zhu 3, Xiekang Wang 4 and Wenjian Tang 1,2Round 1
Reviewer 1 Report
Abstract: What type of modeling is used is not specified.
Page 2, line 68: "....hydrological model...." or "...hydrological modeling..." If "...hydrological model...." then It is necessary to specify which type? analytical, stochastic, statistical, probabilistic, or what?
Page 5, line 196: Acronym HRS should be written also explicitly.
Page 6, line 201: What are the 17 selected parameters?
Page 6, lines 221-222: The authors could apply the standard precipitation index (SPI) for drought, wet and normal classifications in detail. It is necessary how the authors classified these three types in the article?
Page 9, Figure 4: There is random scatter of points. What is the criterion in adapting the two straight lines that pass through the origin?
Page 9, Figure 5: Distinctive meaning of red and blue circles must be explained. Better to put a legend on the figure.
Page 9, Figure 5: Distinctive meaning of red and blue circles must be explained. Better to put a legend on the figure.
Page 10, Figure 6: The authors did not specify which one of these three alternatives, (a), (b) or (c) is reliably used in future studies.
Page 11, lines 344-346: Is it not possible to shift the averages of the simulation to the observation average for better match. What about the standard deviation?
Page 13, lines 367-368: This sentence can be reworded, perhaps as "The correspondence between simulation and observation patterns (see Figure 8). The very word "trend" means some other content in the hydro-meteorology literature as increasing or decreasing trends.
Page 14, line 391: It is better to use "error amounts" instead of "statistics", because statistics imply parameters as mean, standard deviation, skewness, etc. The same is valid for Table 5 on page 15.
Page 17, line 534. Again the use of "trend" is not convenient.
Author Response
Dear Reviewer,
Thanks very much for your reviewing and pertinent comments on our submitted manuscript. We have checked them carefully and find that the quality of our manuscript will be improved significantly if we revised the manuscript according to your suggestions.
The details of the responses to the comments and revisions can be checked as follows and the revisions are marked yellow in the manuscript.
Comment 1: Abstract: What type of modeling is used is not specified.
Response: Thanks very much for your comment. I have revised the sentence in line 20-23.
Revision: Revisited the sentence in line20-23 on page 1 as “Three satellite-based precipitation products, i.e., rain gauge observations (RO), the China Meteorological Assimilation Driving Datasets for the SWAT model (CMADS), and Tropical Rainfall Measuring Mission Multi-satellite (TRMM) products, were used to develop the Soil and Water Assessment Tool (SWAT) model to simulate the streamflow in the Danjiang River Basin (DRB).”. And the sentence “The Soil and Water Assessment Tool (SWAT) was 23 selected to model the streamflow.” was deleted.
Comment 2: Page 2, line 68: "....hydrological model...." or "...hydrological modeling..." If "...hydrological model...." then It is necessary to specify which type? analytical, stochastic, statistical, probabilistic, or what?
Response: Thanks very much for your comment, and we absolutely agree with this comment. It will be better if we specify the type of the hydrological model. The model we used is distributed hydrological model, and we have made some revisions in the related sentences.
Revision: We have revised the "....hydrological model...." or "...hydrological modeling..." as "....distributed hydrological model...." in lines 64-69 (page 2).
Comment 3: Page 5, line 196: Acronym HRS should be written also explicitly.
Response: Thanks very much for your comment. The full name of HRS is "Hydrologic Response Units", and it has been mentioned in line 157 (page 4).
Comment 4: Page 6, line 201: What are the 17 selected parameters?
Response: Thanks very much for your comment. The detailed explanation of the 17 selected parameters were listed in table 6 (page 12).
Revision: We have added supplementary information in the related sentence in line 201 on page 6, and the sentence is revised as "…, 17 parameters were selected (Table 6)."
Comment 5: Page 6, lines 221-222: The authors could apply the standard precipitation index (SPI) for drought, wet and normal classifications in detail. It is necessary how the authors classified these three types in the article?
Response: Thanks very much for your comment, but we think it is better to classify these three types in this article. Because the rainfall varies from years with different precipitation patterns. The classification of the rainfall patterns helps to describe the trend of precipitation more clearly. Moreover, the index we used in this article to classify the rainfall pattern is the index commonly used in China, particularly in Shaanxi Province. The standard of the classification can be found in the reference "Zhang, B.; Xu, X.; Liu, W.; Chen, T. Dynamic changes of soil moisture in loess hilly and gully region under effects of different yearly precipitation patterns. Chinese J. Appl. Ecol. 2008, 19, 1234–1240. ", and the reference has been added in the article. The formulas of the standard are as follows:
Rainy Year: (1)
Drought Year: (2)
Pi was the annual precipitation in ith year, was the average annual precipitation, and ∂ was the mean square error of the average annual precipitation. If Pi>, then it was denoted as a rainy year; if <Pi<, then it was denoted as a normal year; if Pi<, then it was denoted as a drought year.
Revision: We have added the reference after the related sentence in lines 221-223 on page 6.
Comment 6: Page 9, Figure 4: There is random scatter of points. What is the criterion in adapting the two straight lines that pass through the origin?
Response: Thanks very much for your comment, it has been explained in the legend. But we still feel sorry to have not given a clear explanation of what is the criterion in adapting the two straight lines that pass through the origin. That is a straight line with an angle of 45° to the x-axis and y-axis, which is a dividing line, namely, the precipitation products overestimate the rainfall if the point is higher than this line. To make it easier to understand, we added the explanation of that line in the title of Figure 4.
Revision: We added an explanatory sentence ("Note that straight lines that pass through the origin are dividing lines with an angle of 45° to the x-axis, which means the precipitation products overestimated the rainfall if the point is higher than this line.") in the title of Figure 4 (lines 277-279 on page 9).
Comment 7: Page 9, Figure 5: Distinctive meaning of red and blue circles must be explained. Better to put a legend on the figure.
Response: Thanks very much for your comment. Figure 5 (b) and (c) are partial enlargements of Figure 5 (a) and the legends of these pictures have displayed in Figure 5 (a). Namely, the blue, green, and red circles are the cumulative frequencies of daily precipitation intensity for CMADS, TRMM, and Gauge respectively. To make explained the pictures more clearly, we revised the title of Figure 5.
Revision: We revised the title of Figure 5 as " Figure 5. Cumulative frequencies of daily precipitation intensity for Gauge (red points), CMADS (blue points), and TRMM (green points) in the DRB: (a) distribution of all precipitation values, (b) distribution of precipitation values that are < 50 mm, and (c) distribution of precipitation values that are ≥50 mm. " (line 281-284 in page 9). "
Comment 8: Page 10, Figure 6: The authors did not specify which one of these three alternatives, (a), (b), or (c) is reliably used in future studies.
Response: Thanks very much for your comment. It is noticed from Figure 6 that the spatial variation of precipitation at a yearly scale for all sub-basins calculated with precipitation inputs from Gauge seems to reflect more information than that from CMADS and TRMM because the amount of rain gauge far exceeds that of CMADS and TRMM. However, the areal rainfall interpolated from gauges may be distorted, for the gauges observations are point data. The CAMDS and TRMM data are evenly distributed grid data with a resolution of 0.25°, and what they originally reflect are areal rainfall. Therefore, though the gauges observations performed the best in describing the variation of precipitation among these three products, which one will perform the best in simulating runoff is uncertain and which one is reliable to be used in the future should be verified in 3.2.
Revision: We added a section in lines310-316 on page 10 to explain which products are the best in describing the variation of precipitation, like "Though the gauges are able to reflect more information than CMADS and TRMM in describing the variation of precipitation, the areal rainfall interpolated from gauges may be distorted because the gauges observations are point data. But the CMADS and TRMM data are evenly distributed grid data with a resolution of 0.25° and reflect the areal rainfall. Thus, despite the best performance of the rain gauge data in describing watershed areal rainfall among three products, which products perform the best in driving SWAT model to simulate runoff is uncertain.".
Comment 9: Page 11, lines 344-346: Is it not possible to shift the averages of the simulation to the observation average for better match. What about the standard deviation?
Response: Thanks very much for your comment. We have added the standard deviation below the sentence and reworded these sentences.
Revision: We revised the sentence on 355-362 on page 11 as " Besides, the average observed streamflow was smaller than the average simulated runoff throughout the year in the upstream, except in September. The mean standard deviation of the three precipitation products computed at the monthly time scale and averaged over the 8 years considered are 0.28, 1.71, and 4.30 for the Majie Station, Danfeng Station, and Jingziguan Station, respectively. And the mean standard deviation of three products computed at the daily time scale are 0.21, 0.87, and 1.62, respectively."
Comment 10: Page 13, lines 367-368: This sentence can be reworded, perhaps as "The correspondence between simulation and observation patterns (see Figure 8). The very word "trend" means some other content in the hydro-meteorology literature as increasing or decreasing trends.
Response: Thanks very much for your comment. We have found a more appropriate word to replace the word "trend" and reworded this sentence in lines 382-384 (page 14).
Revision: We revised the sentence in lines 379-380 on page 14 as " It is noted from Fig. 8 that there is a positive correlation between the simulated runoff and rainfall."
Comment 11: Page 14, line 391: It is better to use "error amounts" instead of "statistics", because statistics imply parameters as mean, standard deviation, skewness, etc. The same is valid for Table 5 on page 15.
Response: Thanks very much for your comment, we feel sorry to misuse the words and have revised them.
Revision: We revised the title of these tables. For example, the title of Table 5 (lines 360-361 on page 12) as "The pre-calibration performance error amounts of the SWAT model simulated with CMADS, TRMM, and Gauge data on the daily scale." and revised the title of Table 7 (lines 403-403 on page 14) as " The post-calibration performance error amounts of the SWAT model simulated with CMADS, TRMM, and Gauge data on a monthly scale."
Comment 12: Page 17, line 534. Again the use of "trend" is not convenient.
Response: Thanks very much for your comment. We have found a more appropriate word to replace the word "trend" and reworded this sentence in line 552 (page 18).
Revision: We revised the sentence in lines 551-552on page 18 as " However, the rainfall data derived from TRMM and CMADS have a different pattern from the precipitation of Gauge at the daily scale."
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper aims to evaluate the performance of the SWAT model in simulating streamflow discharges in the Dajiang River basin of China by using three sets of precipitation products (RO, CMADS, and TRMM). This study is a valuable addition to the literature of applying satellite-based precipitation products in hydrological modeling. The paper is overall well structured and clearly written. Listed below are some suggestions I have for the authors to consider during the revision.
- The paper needs proofread and there are typos to be fixed. For example, there are four typos just on page 1: Line 17, provide(s); Line 28, date (data); Line 37, pervious (previous); Line 43, are (is).
- Is it reasonable to use the precipitation data interpolated from rain gauges as standard when evaluating the accuracies of the satellite-based precipitation products? The rain-gauge data seems to have the worst performance in the SWAT modeling, which suggests there are significant errors when interpolating the rain gauge data.
- Line 20, "three satellite based precipitation products i.e. rain gauge observations..." is precipitation data observed at rain gauges satellite- based?
- Line 29, change "china" to "China".
- Line 222. what is the commonly used precipitation classification standard ? please describe it.
- Line 250. I thought July is also a peak according to Fig.3.
- Line 316 "..... Gauge with its NSE all below zero". This statement is not correct according to Table 4.
- Line 319, I would change "beyond zero" to "above zero".
- Line 472-479. The statement is not supported by Table 7.
Author Response
Dear Reviewer,
Thanks very much for your reviewing and pertinent comments on our submitted manuscript. We have checked them carefully and find that the quality of our manuscript will be improved significantly if we revised the manuscript according to your suggestions.
The details of the responses to the comments and revisions can be checked as follows and the revisions are marked grey in the manuscript.
Comment 1: The paper needs proofread and there are typos to be fixed. For example, there are four typos just on page 1: Line 17, provide(s); Line 28, date (data); Line 37, pervious (previous); Line 43, are (is).
Response: Thanks very much for your comment. This paper has been proofread carefully and all mistakes have been corrected.
Revision: The typos have been fixed and it has been marked grey.
Comment 2: Is it reasonable to use the precipitation data interpolated from rain gauges as standard when evaluating the accuracies of the satellite-based precipitation products? The rain-gauge data seems to have the worst performance in the SWAT modeling, which suggests there are significant errors when interpolating the rain gauge data.
Response: Thanks very much for your comment. We have considered this question carefully and giving some explanations about this question. The precipitation of rain gauges is interpolated from rain gauges based on the “nearest-distance” principle in the SWAT model. The precipitation of CMADS and TRMM are grid rainfall datasets retrieved from satellites and need to be verified. As the only truth value, precipitation interpolated from gauge is the only dataset that can be used to evaluate the accuracies of the satellite-based precipitation products. In addition, the satellite-based precipitation products have been corrected by ground-based observations, thus they are able to reflect more information when driving the disturbed hydrological models. In summary, though the performance of gauges in simulating runoff is not satisfactory, it can be used to evaluate the performance of satellite-based precipitation products in describing rainfall.
Comment 3: Line 20, "three satellite-based precipitation products i.e. rain gauge observations..." is precipitation data observed at rain gauges satellite-based?
Response: Thanks very much for your comment. The data observed from gauges are not satellite-based data. We have made some revisions to make the accuracy of this sentence.
Revision: We have revised the sentence in line 20-21 on page 1 as " Three precipitation products, i.e., rain gauge observations (RO), the China Meteorological Assimilation Driving Datasets for the SWAT model (CMADS), …".
Comment 4: Line 59, change "china" to "China".
Response: Thanks very much for your comment. We feel sorry to misused the word due to our carelessness and we have revised it.
Revision: We have revised "china" as "China". (line 59 on page 2)
Comment 5: Line 222. what is the commonly used precipitation classification standard? please describe it.
Response: Thanks very much for your comment. To describe the classification standard clearly, we have added a reference that describes this standard in detail. The reference is " Zhang, B.; Xu, X.; Liu, W.; Chen, T. Dynamic changes of soil moisture in loess hilly and gully region under effects of different yearly precipitation patterns. Chinese J. Appl. Ecol. 2008, 19, 1234–1240." In addition, the formulas of the standard are as follows:
Rainy Year: (1)
Drought Year: (2)
Pi was the annual precipitation in the ith year, was the average annual precipitation, and ∂ was the mean square error of the average annual precipitation. If Pi> , then it was denoted as a rainy year; if <Pi< , then it was denoted as a normal year; if Pi< , then it was denoted as a drought year.
Revision: We added a reference after the related sentence in lines 221-223 on page 6.
Comment 6: Line 250. I thought July is also a peak according to Fig.3.
Response: Thanks very much for your comment. We feel sorry to have not described the Fig correctly and have made some revisions.
Revision: We revised the sentence in lines 249-250 on page 8 as "It can be seen from the box plot of Fig. 3 that the precipitation featured three peaks, with the peak values appearing in May, July, and September."
Comment 7: Line 316 ".... Gauge with its NSE all below zero". This statement is not correct according to Table 4.
Response: Thanks very much for your comment. We have made some revisions to ensure the accuracy of words.
Revision: We revised the sentence in lines 323-326 on page 11 as " Besides, the best simulation performance was achieved by CMADS, whose NSE was 0.74 upstream and 0.63 downstream (Jingziguan Station), while the worst simulation effect occurred in Gauge with its NSE almost all below zero (only above zero in the Majie Station).".
Comment 8: Line 319, I would change "beyond zero" to "above zero".
Response: Thanks very much for your comment. We feel sorry to misused the words and have made some revisions to ensure the accuracy of words.
Revision: We revised the sentence in lines 328-331 on pages 11 as "The PBIAS of the CMADS model and TRMM model were both above zero in up-stream and below zero in the midstream and downstream, …".
Comment 9: Line 472-479. The statement is not supported by Table 7.
Response: Thanks very much for your comment. We have made some revision to make the statement can supported Table 7.
Revision: We revised the sentence in lines 484-489 on pages 16-17 as " Pre-calibration results showed that the CMADS and TRMM, and gauge data were all reliable to estimate runoffs on the monthly scale at Majie Station and Jingziguan Station, while they performed unsatisfactory in simulating streamflow at Danfeng Station. The performances of Gauges in estimating runoff on the monthly scale in the middle stream and downstream were both unreliable, only its performances in runoff simulation at the Majie Station was satisfactory. The performances of that on the daily scale, however, were all unsatisfactory."
Author Response File: Author Response.pdf
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
