UAV Multispectral Imagery Combined with the FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain

Round 1

Reviewer 1 Report

Authors have provided partially satisfactory responses to my comments and concerns, but did not make a serious effort to address some of the important comments. In reviewing the revised manuscript, I have more questions and comments.

 

round 1, Point 1: For example, it is not clear how Equation 5 is implemented. Mainly, how “Dr” is determined. Is it from daily water balance model etc?

The did not even respond to this except that they indicated an interpolation between measured points. The equations (R1, R2,3) are not shown in the responses or in the revised document.

 

It is is till not clear why they interpolated the observed while they have an equation that was presented for conducting daily water balance or not clear why they did not use equation 12 to determine SW and SW depletion?

 

round 1, Points 5 and 6: While the promise of UAV is great in terms of high spatiotemporal resolution, it will be useful to also discuss the challenges associated with scaling to larger areas and also the management aspect of handling the volume of such high frequency data for operational applications.

 

it is not clear why they presented a literature review of how soil moisture estimation is made while my question was why they did not conduct a daily water balance through all these integration of soil moisture, VI and weather datasets , which is similar to the equation presented. Soil measurement is good for model validation, but for regular monitoring using UAV, we should demonstrate how UAV can help improve our spatial coverage with limited in situ measurement.

 

Similarly,  I was asking for challenges of using UAV for regular monitoring. Instead I was directed to same material that talks about the strengths of UAV compared to satellites.

 

In addition to being not very responsive to the comments and concerns, there are a few more issues. Some minor and some more serious:

 

Line 334: reference to a non-existing equation 13

How are TCARI and RDVI calculated? These are not common indices. One has to read another paper to know such important parameters that are used in this manuscript. 

-there was no attempt or discussion on why LAI was not calculated from NDVI. From soil moisture to LAI, the study is focusing on measurement rather than remote sensing approaches.  Interestingly, Ks was determined from images but not LAI. One would think there is more literature support to estimate LAI than Ks from VIs 

Equation 5 wrongly referenced on Line 399

Figure 5: Ks-FAo: is that new? Or same as Ks-Tab?

Line 324: vapor pressure gradient (VPG), respectively [38]. Still points to the wrong reference.

Line 334:  refers to Equation(13) : There is no Equation 13!

Eqn 12 is not defined well. What is the need for "+/-" operation before dW?  What happens if you simply use “ …"x - dW"”?

Along with that, it seems dW is assumed zero for lack of capillary rise? But that is not the definition of dW in water balance equation. This lets me question if the water balance is set up correctly. Not clear what the final water balance equation was.  Was it simply ET = P  + I  because all the rest (Ro, DP, and  dW are assumed 0.0? This has to be clarified.

 

Similarly, equation 5 is not well defined (about ks). What are the units for each of the parameters. What happens to Ks when Dr is 0.0 . What are the ranges of Dr? from 0 to 0.35TAW? How are field capacity and wilting point defined? What are their values for the study sites? A table for model parameters may be required.

 

Figure 6 is still problematic, plotting 3 to 5 days accumulation in one graph which could inflate R2 values. Why not show one time scale, either 3 day or 5-day. This is like plotting, daily, 10-day and yearly in one chart which would inflate r2 values.

 

Conclusion:

Line 557:  it appears that the manuscript considers water balance ET as observed ET. That is still modeled ET and should not be considered as observed.

 

Again, the study is an extension of your previous paper, but methods are not as clear as they could be and results are pretty much expected, but the conclusions appear to be making far reaching assertion on usefulness of UAV for irrigation scheduling. While the study depended on measured soil moisture and LAI instead of UAV measured parameters, it is a bit too strong to recommend/promise UAV for regular monitoring. Outlining the next steps in terms of model parameterization would be more appropriate.

 

 

 

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-598317). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

 

Point 1: It is till not clear why they interpolated the observed while they have an equation that was presented for conducting daily water balance or not clear why they did not use equation 12 to determine SW and SW depletion?

 

Point 2: round 1, Point 1: For example, it is not clear how Equation 5 is implemented. Mainly, how “Dr” is determined. Is it from daily water balance model etc?

They did not even respond to this except that they indicated an interpolation between measured points. The equations (R1, R2,3) are not shown in the responses or in the revised document.

 

Point 3: Similarly, equation 5 is not well defined (about ks). What are the units for each of the parameters. What happens to Ks when Dr is 0.0. What are the ranges of Dr? from 0 to 0.35TAW? How are field capacity and wilting point defined? What are their values for the study sites? A table for model parameters may be required.

 

Response 1: Thank you, we employed the Equation (5) (from FAO 33) which directly use soil water content to calculate Ks. The reason why we didn't use the water balance method to calculate daily Dr and then obtained the Ks is that we get some values of ET₀ less than 0 due to unstable weather station data when we calculate the ET. It may bring errors with the daily water balance method. In equation (5) (from FAO 33), we use soil water content directly to calculate Ks. When soil water content lower than the θj (the threshold water content) crop begins to reach the stress period (Ks<1). The Equation (5) is an equivalent expression to the equation in the previous manuscript which in terms of depletion [1]. Reference [2] also employ equation (5).

 

Response 2: Thank you, change has been made in Line 270-271.

 

Response 3: Thank you, change has been made in Line 272-281.

Water content in the root zone can be expressed by root zone depletion, Dr, (i.e., water shortage relative to the field capacity). At field capacity, the Dr =0, Ks=1. When soil water is extracted by evapotranspiration, the depletion increases and stress will be induced when Dr becomes equal to RAW (0.35 TAW in this manuscript). In the equation (5), which use the water content to calculate Ks. When the water content is lower than , crop begins to reach the stress period (Ks<1). and if less than  crop do not absorb water from root zone (Ks = 0). The parameters for FAO-56 dual crop coefficient method listed in Table 2.

 

Point 4: round 1, Points 5 and 6: While the promise of UAV is great in terms of high spatiotemporal resolution, it will be useful to also discuss the challenges associated with scaling to larger areas and also the management aspect of handling the volume of such high frequency data for operational applications.  It is not clear why they presented a literature review of how soil moisture estimation is made while my question was why they did not conduct a daily water balance through all these integration of soil moisture, VI and weather datasets, which is similar to the equation presented. Soil measurement is good for model validation, but for regular monitoring using UAV, we should demonstrate how UAV can help improve our spatial coverage with limited in situ measurement.

 

Response 5: Thank you, sorry about that we misunderstand you valuable comments. We have added the discussion about the challenges of UAV in Line 612-616. The goal of this manuscript is that we want directly to estimate the ET in water stress condition with UAV. We revised the manuscript and discussed the significance that accurately estimated the ET under water stress condition by simple VIs method. The soil moisture data and water balance equation was used to calculate cumulated ET during the 6–29 August 2017 and evaluate ability of ET determination through UAV. With the result of this manuscript, we can obtain the ET only with meteorological data and VIs. Conducting a daily water balance through all these integration of soil moisture, VI and weather datasets like the reference [2,3] is not the scope of this manuscript.

 

Point 6: Line 334: reference to a non-existing equation 13.

 

Response 7: Thank you, change has been made in Line 343-349.

 

Point 8: How are TCARI and RDVI calculated? These are not common indices. One has to read another paper to know such important parameters that are used in this manuscript.

 

Response 8: Thank you, change has been made in Line 313 and 314.

 

Point 9: There was no attempt or discussion on why LAI was not calculated from NDVI. From soil moisture to LAI, the study is focusing on measurement rather than remote sensing approaches.  Interestingly, Ks was determined from images but not LAI. One would think there is more literature support to estimate LAI than Ks from VIs

 

Response 9: Thank you, the observed LAI is used to modify Kcb (Kcb-Tab) and evaluate the effects of water deficit on crops. The NDVI is used to calculate Kcb-NDVI. And the observed soil moisture is used for Ks-FAO calculation (equation 5 and 6). In addition to the Ks-FAO, we also estimated the Ks-CWSI and Ks-ratio based on canopy temperature. By discussing the shortcomings of previous calculating method of Ks and selecting the optimal water stress index for retrieving, we obtained the spatial and temporal distribution of Ks through VIs (i.e. equation (15)) at field scale. Finally, we can obtain Kcb and Ks synchronously through UAV, which may be help for ET estimation. Therefore, using VIs to estimate LAI is not the focus and content of this manuscript. 

 

Point 10: Equation 5 wrongly referenced on Line 401

 

Response 10: Thank you, change has been made in Line 269.

 

Point 11: Figure 5: Ks-FAO: is that new? Or same as Ks-Tab?

 

Response 11: Thank you, change has been made in Figure 7 (Line 461).

 

Point 12: Line 324: vapor pressure gradient (VPG), respectively [38]. Still points to the wrong reference.

 

Response 12: Thank you, change has been made in Line 341. Reference [34] in the manuscript.

 

Point 13: Line 334: refers to Equation (13) : There is no Equation 13!

 

Response 13: Thank you, change has been made in Line 343-349.

 

Point 14: Eq 12 is not defined well. What is the need for "+/-" operation before dW? What happens if you simply use “…"x - dW"”?

 

Response 14: Thank you, change has been made in Line 370

 

Point 15: Along with that, it seems dW is assumed zero for lack of capillary rise? But that is not the definition of dW in water balance equation. This lets me question if the water balance is set up correctly. Not clear what the final water balance equation was.  Was it simply ET = P + I  because all the rest (Ro, DP, and  dW are assumed 0.0? This has to be clarified.

Response 15: Thank you, change has been made in Line 371-378.

 

Point 16: Figure 6 is still problematic, plotting 3 to 5 days accumulation in one graph which could inflate R2 values. Why not show one time scale, either 3 day or 5-day. This is like plotting, daily, 10-day and yearly in one chart which would inflate r2 values.

 

Response 16: Thank you, because reviewers question the way of validation. We have change the method. Studies [4,5] have suggested that modified dual crop coefficient approach was suitable for calculating the actual daily evapotranspiration of the main crops (including maize) in the North China Plain. Thus, the simulated daily ET of the maize derived from the 2 Ks methods and NDVI-based Kcb methods (ET-CWSI and ET-ratio) were compared with the values obtained from modified FAO-56 Kc method.

 

Point 17: Line 557: it appears that the manuscript considers water balance ET as observed ET. That is still modeled ET and should not be considered as observed.

 

Response 17: Thank you, we referred to the water balance method in reference [6], and we compare cumulative evapotranspiration (CET) obtained by VIs method with water balance to evaluate its ability to determine water consumption. According this reference, the comparison method has certain accuracies.

 

Point 18: Again, the study is an extension of your previous paper, but methods are not as clear as they could be and results are pretty much expected, but the conclusions appear to be making far reaching assertion on usefulness of UAV for irrigation scheduling. While the study depended on measured soil moisture and LAI instead of UAV measured parameters, it is a bit too strong to recommend/promise UAV for regular monitoring. Outlining the next steps in terms of model parameterization would be more appropriate.

 

Response 18: Thank you, according to the reviewers’ comments, we changed validation method. The simulated daily ET of the maize derived from the 2 Ks methods and NDVI-based Kcb methods (ET-CWSI and ET-ratio) were compared with the values obtained from modified FAO-56 dual crop coefficient method Equations (1-6). The ET-CWSI showed a better agreement with the modified FAO-56 dual crop coefficient method. The suitable water stress is important for Ks estimation through multispectral UAV. Due to lacking observed soil moisture data and in order to evaluate the ability of ET determination by UAV, we refer to the reference [7] and added a table which compare cumulative evapotranspiration obtained by VIs method with water balance approach. The difference between the cumulative ET calculated by VIs method and water balance approach were 2.6 mm, 8.9 mm, and 5 mm, during 6–29 August 2017, respectively. Note that the retrieved ET were familiar with the crop water consumption. Thus, based on which, regular monitoring by UAV has a certain guiding significance. Future experiments will incorporate ground validation (eddy covariance or lysimeter) of ET to provide an independent assessment of model accuracy and use more convenient and reliable water stress indices to evaluate crop stress and quantify crop evapotranspiration in longer crop growth period.

 

 Thank you again for your valuable comments！

 

 

 

Reference

 

Jensen, M.; Allen, R. Evaporation, evapotranspiration, and irrigation water requirements; 2016; Vol. 2016, pp. 1-744. Olivera-Guerra, L.; Merlin, O.; Er-Raki, S.; Khabba, S.; Escorihuela, M.J. Estimating the water budget components of irrigated crops: Combining the FAO-56 dual crop coefficient with surface temperature and vegetation index data. Agricultural Water Management 2018, 208, doi:10.1016/j.agwat.2018.06.014. Consoli, S.; Vanella, D. Mapping crop evapotranspiration by integrating vegetation indices into a soil water balance model. Agricultural Water Management 2014, 143, 71–81, doi:10.1016/j.agwat.2014.06.012. Feng, Y.; Cui, N.; Gong, D.; Wang, H.; Hao, W.; Mei, X. Estimating rainfed spring maize evapotranspiration using modified dual crop coefficient approach based on leaf area index. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2016, 32, 90-98, doi:10.11975/j.issn.1002-6819.2016.09.013. Ding, R.; Kang, S.; Zhang, Y.; Hao, X.; Tong, L.; Du, T. Partitioning evapotranspiration into soil evaporation and transpiration using a modified dual crop coefficient model in irrigated maize field with ground-mulching. Agricultural Water Management 2013, 127, 85-96, doi:https://doi.org/10.1016/j.agwat.2013.05.018. Zeleke, K.; Wade, L. Evapotranspiration Estimation Using Soil Water Balance, Weather and Crop Data. 2012; 10.5772/17489. Bausch, W.; Trout, T.; Buchleiter, G. Evapotranspiration adjustments for deficit-irrigated corn using canopy temperature: A concept. Irrigation and Drainage 2011, 60, 682-693, doi:10.1002/ird.601.

 

Reviewer 2 Report

Line 187. Regions can be confusing change by zones

Line 202. Do all zones have the same texture?. A table with texture could be interesting.

Line 212. dual crop coefficient method instead of "crop coefficient"

Line 213. single crop coefficient instead of "tabular method"

Line 231. Why not measured at solar noon? All measures were measured in the same hour?. The temperature can change quickly

Line 250. Where are ETo and Ta values taken? agrometeorological station on the field or next?. please add

Line 302. 4.7 cm is this the pixel resolution? please add

In figure 6 In ET the measure of Ke must be indicated in relation to the water applied and the cover of the crop, has this been taken into account? Explain, please.

Line 471 It should be explained more deeply why the CV is 40% in TR2

Eddy covariance systems should have been used.

basal crop coefficients were adjusted for that crop in the area? or have been considered general kcb of FAO 56?. is it very important since a local adjustment is necessary.

 

 

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-598317). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

 

Point 1: Line 187. Regions can be confusing change by zones

 

Response 1: Thank you, change has been made in Line 156, 157, 161,163, 174 and 175.

 

Point 2: Line 202. Do all zones have the same texture?. A table with texture could be interesting.

 

Response 2: Thank you, we employed the average soil characteristics in the field, and the heterogeneity was neglected in the experimental treatment. The information of soil characteristics are listed in Table 2.

 

Point 3: Line 212. dual crop coefficient method instead of "crop coefficient"

 

Response 3: Thank you, change has been made in Line 181.

 

Point 4: Line 213. single crop coefficient instead of "tabular method"

 

Response 4: Thank you, change has been made in Line 183.

 

Point 5: Line 231. Why not measured at solar noon? All measures were measured in the same hour? The temperature can change quickly

 

Response 5: Thank you, sorry for the inexplicit statement. All the temperature measurements in one experiment were at the solar noon and taken about 20 min, not from 11:00 to 13:00. 11:00-13:00 was the time window we could measure. Change has been made in Line 212.

 

Point 6: Line 250. Where are ETo and Ta values taken? agrometeorological station on the field or next?. please add

 

Response 6: Thank you, change has been made in Line 199-220 and 381-383.

 

 

Point 7: Line 302. 4.7 cm is this the pixel resolution? please add

 

Response 7: Thank you, change has been made in Line 304.

 

Point 8: In figure 6 In ET the measure of Ke must be indicated in relation to the water applied and the cover of the crop, has this been taken into account? Explain, please.

 

Response 8: Thank you, change has been made in Line 469-471.

 

Point 9:  Line 471 It should be explained more deeply why the CV is 40% in TR2

 

Response 9: Thank you, the content about the CV of ET have been rewritten, and explain the reason about the difference of CV in different treatments. Changes have been made in Line 519-528.

 

Point 11: Eddy covariance systems should have been used and basal crop coefficients were adjusted for that crop in the area? or have been considered general kcb of FAO 56?. is it very important since a local adjustment is necessary.

 

Response 11: Thank you, there are many articles using eddy covariance to obtain accurate daily ET values. However, their popularization and application are limited by the complex design technology and high cost of various sensors. Eddy covariance are mostly used to detect large areas of crops and require uniform underlying surfaces [1]. But our sample areas are small and close with each other. Meanwhile, our experiment is divided into three irrigation levels. Instead, reference [2,3] modified basal crop coefficient method with LAI through eddy covariance systems near the experiment site, which illustrated the modified dual crop coefficient method could estimate maize ET accurately on the North China. Therefore we use the ET calculated by modified dual crop coefficient method to compare with the ET-CWSI and ET-ratio.

 

Thank you again for your valuable comments!

 

 

 

Reference

 

Culf, A.D.; Foken, T.; Gash, J.H.C. The Energy Balance Closure Problem; 2004. Feng, Y.; Cui, N.; Gong, D.; Wang, H.; Hao, W.; Mei, X. Estimating rainfed spring maize evapotranspiration using modified dual crop coefficient approach based on leaf area index. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2016, 32, 90-98, doi:10.11975/j.issn.1002-6819.2016.09.013. Ding, R.; Kang, S.; Zhang, Y.; Hao, X.; Tong, L.; Du, T. Partitioning evapotranspiration into soil evaporation and transpiration using a modified dual crop coefficient model in irrigated maize field with ground-mulching. Agricultural Water Management 2013, 127, 85-96, doi:https://doi.org/10.1016/j.agwat.2013.05.018.

 

Reviewer 3 Report

L61 include some references to support the acceptance of the FAO method.

L64-65 Rephrase “Kcb and Ks, as the coefficients truly reflect crop growths status, they have attracted wide attention.”

L67 Include a reference to this method.

L82 remove “,” after “this paper”.

L83-84 Rephrase “When the soil water content is less than the level of maximum allowable depletion, crop stress coefficient (Ks) should be considered by irrigation manager.”

L88 include a reference for ” However, water stress is very common in practical agricultural activities”.

L97 include another two references since the sentence started with “Studies”.

L208 Table 1, explain what are the values in parenthesis under late vegetative (they seem to be dates).

L218 Figure 2. Define “CET”.

L235 Include the reference to the manufacturer error, is this brightness temperature?

L235-236 Is this the range of the observed temperatures in the field? Probably there is a typo and it should read 32-60 C.

L237 Why was this emissivity value chosen?

L242 Properly define the index as leaf area index (LAI).

L251 Add a reference to the wide adoption of the method.

L269-273 Kcb,table doesn’t appear on eq. 3 nor 4. Why was the value of 0.7 chosen? Or how was this parametrization derived?

L287 missing “s” in pixels.

L292 Which Gray plate? Was this used as a calibration target for the multispectral camera? This is specified a few paragraphs after it was first introduced. Either remove the first reference or explain it before L304. How was the calibration done (empirical line?)?

L311 overpasses.

L330 Which VIs?

L396 Table 2, convert to a plot similar to Figure 4 for easy visualization and comparison of the different variables under the treatments. Also, expand on the discussion of these differences between estimation methods and its impact to ET.

L413 “were less water stressed than”

L483-489 This information is already contained in the introduction and should not be part of the discussion.

L490-491, provide references for the contrasting studies.

L497 Did the authors meant finally?

L504 Why was this alternative not explored in the paper given that it was a probable cause of divergence? (the use of SAVI and how it changed the results).

L513-519 Remove this part or integrate into the introduction.

The low coefficient of determination could also come from the relatively low number of UAV flights, this should also be addressed in the discussion.

Consider including an additional figure that compares the ET derived from the different methods and a small discussion on the water used by each of them.

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-598317). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

Because we have made a comprehensive revision of the structure of each part of the manuscript based on your comments and those of other reviewers, some of the false sentence that you pointed have been deleted. Hope you can understand.

 

Point 1: L61 include some references to support the acceptance of the FAO method.

 

Response 1: Thank you, change has been made in Line 62.

 

Point 2: L64-65 Rephrase “Kcb and Ks, as the coefficients truly reflect crop growths status, they have attracted wide attention.”

 

Response 2: Thank you, which has been deleted.

 

Point 3: L67 Include a reference to this method.

 

Response 3: Thank you, this part of the method is introduced separately in the new introduction. And we added references [5] and [8] in Line 61and 71.

 

Point 4: L82 remove “,” after “this paper”.

 

Response 4: Thank you, which has been deleted.

 

Point 5: L83-84 Rephrase “When the soil water content is less than the level of maximum allowable depletion, crop stress coefficient (Ks) should be considered by irrigation manager.”

 

Response 5: Thank you, which has been deleted.

 

Point 6: L88 include a reference for ” However, water stress is very common in practical agricultural activities”.

 

Response 6: Thank you, which has been deleted.

 

 

Point 7: L83-84 Rephrase “When the soil water content is less than the level of maximum allowable depletion, crop stress coefficient (Ks) should be considered by irrigation manager.”

 

Response 7: Thank you, which has been deleted.

 

Point 8: L97 include another two references since the sentence started with “Studies”.

 

Response 8: Thank you, change has been made in Line 177-178.

 

Point 9: L208 Table 1, explain what are the values in parenthesis under late vegetative (they seem to be dates).

 

Response 9: Thank you, they are dates of different grown stages, change has been made in Line178-179.

 

Point 10: L218 Figure 2. Define “CET”.

 

Response 10: Thank you, change has been made in Line 192.

 

Point 11: L235-236 Is this the range of the observed temperatures in the field? Probably there is a typo and it should read 32-60 C.

 

Response 11: Thank you, the temperature measurement range of the instrument is -32–600 °C, and the change has been made in Line 218.

 

Point 12: L235 Include the reference to the manufacturer error, is this brightness temperature?

 

Response 12: Thank you, which mean temperature (℃). More specific information can be found in: https://www.raytek-direct.com/product/raytek-rayst81-st-pro-plus-portable-ir-thermometer-with-50-1-optics-32-to-760c.

 

Point 13: L237, Why was this emissivity value chosen?

 

Response 13: Thank you, when the emissivity of the infrared thermometer is the same as the emissivity of the measured object, the indication value is the real temperature of the measured object. A canopy was seen by an infrared thermometer includes leaves and cavities formed by the canopy architecture that closely approximate blackbodies (emissivity = 1), but true emissivity of plant leaves are actually less than 1. Idso et al. [1]  demonstrated that emissivity of plant leaves are about 0.97-0.98 , and the reference has been added in Line 217.

 

Point 14: L251 Add a reference to the wide adoption of the method.

 

 

Response 14: Thank you, change has been made in Line 235.

 

Point 15: L269-273 Kcb, table doesn’t appear on eq. 3 nor 4. Why was the value of 0.7 chosen? Or how was this parametrization derived?

 

Response 15: Thank you, change has been made in Line 259-263, and the value of 0.7 is from the reference [2,3], and listed in Table 2.

 

Point 16: L287 missing “s” in pixels.

 

Response 16: Thank you, change has been made in Line 287.

 

Point 17: L292 Which Gray plate? Was this used as a calibration target for the multispectral camera? This is specified a few paragraphs after it was first introduced. Either remove the first reference or explain it before L304. How was the calibration done (empirical line?)?

 

Response 17: Thank you, change has been made in Line 293-295. The original data acquired by the multispectral camera is the DN value of the ground target, and the data is stored in the raw format. The data acquired by the camera sensor is arranged in Bayer mode, i.e. the four adjacent pixels of the camera sensor are three different channels, respectively. The original data is pre-processed by pix 4d, and the data of three channels are separated. Then, pix 4d uses relative measurement method (i.e. under the same observation conditions, the DN values of ground targets and whiteboard are recorded separately.) to calibrate the data through the gray plates. All image processing is done in pix4d, including radiation correction.

 

Point 18: L311 overpasses.

 

Response 18: Thank you, change has been made in Line 311.

 

Point 19: L330 Which VIs?

 

Response 19: Thank you, change has been made in Line 343.

 

Point 20: L396 Table 2, convert to a plot similar to Figure 4 for easy visualization and comparison of the different variables under the treatments. Also, expand on the discussion of these differences between estimation methods and its impact to ET.

 

Response 20: Thank you, change has been made in Line 433 (Figure 6). Because change the method of Kcb-NDVI equation (10) and (11), the new Kcb-NDVI showed closely tracked modified Kcb-Tab.

 

Point 21: L413 “were less water stressed than”

 

Response 21: Thank you, change has been made in Line 451.

 

Point 22: L483-489 this information is already contained in the introduction and should not be part of the discussion.

 

Response 22: Thank you, we have reorganized the structure of the introduction and discussion and we delete the repetitive information in the introduction.

 

Point 23: L490-491, provide references for the contrasting studies.

 

Response 23: Thank you, change has been made Line 608. Reference [61] and [62] were added.

 

Point 24: L497 Did the authors meant finally?

 

Response 24: Thank you, which has been deleted.

 

Point 24: L504 Why was this alternative not explored in the paper given that it was a probable cause of divergence? (the use of SAVI and how it changed the results).

 

Response 24: Thank you, we have change the retrieve relationship between VI and Kcb. Due to the water stress and our planting density, the vegetation cover didn’t reach to 100% cover, and reference [4-6] showed that Kcb can be estimated from fractional vegetation cover. Thus, we use the Equation (10) and (11) to calculate Kcb and Figure 6 showed stronger similarity between the Kcb-NDVI and Kcb-Tab.

 

Point 24: L513-519 Remove this part or integrate into the introduction.

 

Response 24: Thank you, change has been made.

 

Point 24: The low coefficient of determination could also come from the relatively low number of UAV flights, this should also be addressed in the discussion. And consider including an additional figure that compares the ET derived from the different methods and a small discussion on the water used by each of them.

 

Response 24: Thank you, because reviewers question the way of validation. So we change the method. Several studies [7,8] have suggested that crop coefficient method proposed in FAO-56 was suitable for calculating the actual daily evapotranspiration of the main crops (winter wheat and summer maize) in the North China Plain, and modified dual crop coefficient approach [2,3]. Thus, the simulated daily ET of the maize derived from the 2 Ks methods and NDVI-based Kcb methods (ET-CWSI and ET-ratio) were compared with the values obtained from standard FAO-56. Because we And we have compared cumulative evapotranspiration (CET) obtained by VIs method with water balance to evaluate the ability of UAV to determine irrigation amount. Change has been made in Line 494-506 (Table 5). The difference between the cumulative calculated by VIs method and measured ET were 2.6 mm, 8.9 mm, and 5 mm, respectively. Note that the retrieved ET were familiar with the crop water consumption. Of course, future experiments will incorporate ground validation (eddy covariance or lysimeter) of ET to provide an independent assessment of model accuracy.

 

Thank you for your valuable comments about this manuscript!

 

 

Reference

 

Idso, S.B.; Jackson, R.D.; Ehrler, W.L.; Mitchell, S.T.J.E.V. A Method for Determination of Infrared Emittance of Leaves. Ecology 1969, 899-902. Ding, R.; Kang, S.; Zhang, Y.; Hao, X.; Tong, L.; Du, T. Partitioning evapotranspiration into soil evaporation and transpiration using a modified dual crop coefficient model in irrigated maize field with ground-mulching. Agricultural Water Management 2013, 127, 85-96, doi:https://doi.org/10.1016/j.agwat.2013.05.018. Feng, Y.; Cui, N.; Gong, D.; Wang, H.; Hao, W.; Mei, X. Estimating rainfed spring maize evapotranspiration using modified dual crop coefficient approach based on leaf area index. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2016, 32, 90-98, doi:10.11975/j.issn.1002-6819.2016.09.013. Johnson, L.F.; Trout, T.J. Satellite NDVI Assisted Monitoring of Vegetable Crop Evapotranspiration in California's San Joaquin Valley. Remote Sensing 2012, 4, 439-455, doi:10.3390/rs4020439. Trout, T.; Johnson, L.; Gartung, J. Remote Sensing of Canopy Cover in Horticultural Crops. HortScience 2008, 43, doi:10.21273/HORTSCI.43.2.333. Carlson, T.N.; Ripley, D.A. On the relation between NDVI, fractional vegetation cover, and leaf area index. Remote Sensing of Environment 1997, 62, 241-252, doi:10.1016/s0034-4257(97)00104-1. Liu, Y.; Luo, Y. A consolidated evaluation of the FAO-56 dual crop coefficient approach using the lysimeter data in the North China Plain. Agricultural Water Management 2010, 97, 31-40, doi:https://doi.org/10.1016/j.agwat.2009.07.003. Zhang, X.; Chen, S.; Sun, H.; Shao, L.; Wang, Y. Changes in evapotranspiration over irrigated winter wheat and maize in North China Plain over three decades. Agricultural Water Management 2011, 98, 1097-1104, doi:https://doi.org/10.1016/j.agwat.2011.02.003.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Critique:

I would like to congratulate the authors on the substantial improvement to the second version of the manuscript. The authors base their study on a previous field campaign to estimate crop coefficients by using multispectral UAV imagery. Through their study, it is clear that using remotely sensed information is a good alternative to derive these crop coefficients. However, the number of observations during the growing seasons area limitation to the study as well as an apparent lack of validation information. It would be interesting to conduct another study with more UAV flights to see if the proposed methodology works as well during the early stage development. Furthermore, a comparison with satellite derived crop coefficients could enhance the discussion.

Major comments:

On line 471, the authors say that the remote sensing model was evaluated using ET-FAO data, does this mean that the evaluation was just a comparison to the tabular method? Or was there measured ET from an eddy covariance tower/lysimeter. If there were no actual observations, please include a comparison on the ET rates to similar studies.

On line 596 “For instance, Landsat and SPOT have spatial resolution of 30 and 20 m, respectively.”  The authors only mention Landsat and SPOT, however, there are better satellites in use. An example is Sentinel 2 with a ~5 day retrieval period and 10 m resolution on the optical bands. Also, there are CubeSats with near daily acquisition time and 3 m spatial resolution. The authors should justify on the focus on Landsat and SPOT or include some references and discussion of new observation platforms.

On lines 598-599, the authors suggest that UAVs can monitor information up to the leaf scale. Is this feasible and economically viable for farmers? What would be the possible difficulties if so?

L599-602 “From the ET maps (Figure 9), we can observed the evapotranspiration of crops varied even with the same treatment. Table 4 showed the mean and CV values of different treatments due to the different soil texture and soil heterogeneity in the field.” The authors need to discuss the implications of this information rather than just repeating it from the results.

Minor comments:

What do the authors mean by “data conversion” on L126. Also, note that thermal imagery has the added burden of calibration which severely affects the absolute accuracy of the LST and can introduce other artifacts on the image.

On line 225 it was mentioned that a cubic spline was used to derive daily LAI and canopy height, can the authors explain why was a cubic spline used as the interpolation method? How can these parameters affect regions of high heterogeneity? Were they modified over those areas to be representative?

Please mention the method to correct reflectivity data from the grey plate on line 288.

On line 336, there is a mention to upper and lower canopy temperatures, how are these limits computed? Do they come from the sample ensemble? Are they measured at the same time? Or are they the minimum and maximum daily canopy temperatures?

On lines 345 and 346, there seems to be two equations for the same variable, but the text only mentions the equation on line 346. Can the authors clarify which equation was used? The 0 and 1 values for CWSI seem to be flipped for the same intervals.

On lines 549 to 556, much of this information should not be part of the discussion. The authors should explain what are the effects of each method (direct or indirect) on the coefficients rather than provide an introduction to them. PET can be estimated, the difficulty is on measuring actual ET.

On lines 127 to 129, the authors state that “For example, due to the image resolution obtained by the two sensors is different, irrigation manager need to downscale the spatial resolution of multispectral map to match the scale of thermal imagery.” However, usually, the lower resolution sensor is the thermal one, furthermore, there is not a clear agreement on the required spatial resolution and how and when it begins to affect predictions based on UAV observations. And, most importantly, the biggest effect is not only on the pixel mixing due to the downscale, but on the georectification and coregistration of the images.

L66 “However, the method is”

L68 “ due to inconsistent crop growth”

L72 “large irrigated fields”

L82 Please add a few examples of these new technologies with their references.

L89 “when the crop completely covers”

L89 “Many studies have”

L101 “It is necessary to use a new indicator”

L103 “canopy temperature is”

L113 “that using an appropriate”

L119-L120 Rephrase “which can be greatly constrained by labor intensive and time consuming”

L121 “estimate ET”

L124 “CWSI, and water deficit index”

L126-127 “obstacle to estimate ET with UAV systems”

L129 “the Ks and Kcb may influence”

L135 Rephrase “using one multispectral UAV should be explored.” The UAV is not multispectral but the camera/sensor attached to it is.

L136 “This study aims to estimate more”

L141-143 “through the relation between water stress index and VIs, to eliminate discrepancies caused by using data at different scales and obtain high resolution spatiotemporal ET maps.” The author should include the achieve resolution(s) in parenthesis, 10 cm for example.

L164 “MIK-2000H flow meters” provide the accuracy of the flow meters.

L214 Please specify that these ranges are for the thermal instrument.

L253 “the canopy height and LAI were used to modify the dynamic Kcb.” The use of canopy height rather than h is a personal preference as it makes the text more readable and is not an overly long word.

L258 What is the value of k and where is it on equations 3-4?

L324 “has been obtained through” Or add another method to obtain Kcb.

L327-329 “The two different Ks obtained from CSI and Tc ratio were derived from the handheld infrared thermometer, with daily values taken between 11:00 and 13:00 (local time), which are approximate times of peak stress.”

L351 How is the variable Tc ratio estimated? What is this ratio?

L354 “have suggested that the modified dual”

L357 “from the modified FAO-56”

L390 “the highest Tc in TRT 2 reached to 34°C.”

Figure 4 Modify the x-title to include the year.

L445-446 Can you elaborate on this? “It’s probable that even a well-watered crop could have a high canopy temperature because of irrigation followed by very hot day.”

L473-475 “It can be clearly seen from Figure 8 that the two different water stress indices model performed their evaluation results from the coefficient of determination (R2), root mean square error (RMSE) and index of agreement (d).” Rephrase, it is not clear what the authors meant with this sentence. Is it that the two stress indices were evaluated based on the previously mentioned statistics?

L476 “Figure 8 (a) and (b), is greater than 0.9, which”

L476 “ET based on water stress has a strong fit” A perfect fit would be for a value of one.

L478 Please include in parenthesis a NRMSE value which will give readers a better picture if this is a low or high value (normalize by the mean).

L500 “The different soil texture, and soil heterogeneity lead”

L506 Rephrase “which illustrated that the prolonged water stress in anomaly distinctly amplified spatial fluctuations in field soil heterogeneity via its influences on maize evapotranspiration condition.”

L519 “were calculated by”

L521 “As mentioned above, crop coefficients was derived”

L524 “the total water consumption of the three treatments”

L526 “ET calculated by the Vis”

L527 Rephrase “Note that the retrieved ET were familiar with the crop water consumption.” It is not clear what the authors mean with “familiar”

L533 “of models for ET assessment.”

L533-534 Rephrase “Models include surface energy balance model and strictly empirical VI model.”

L534 “Though the application of surface energy balance models are able to” Who is able? Who is the subject of this sentence?

L535 “accuracy. Deficiencies” This is a separate sentence.

L541-542 “Multispectral VIs calculated from canopy reflectance can be used to simulate real-time Kcb.”

L543 “help of UAVs”

L543 Which studies? Add the appropriate references.

L555-556 rephrase “In many situation, it's impossible to get the data of PET and actual ET when lack of

556 sophisticated and costly instruments such as eddy covariance and lysimeter.”

L556 “Diarra et al. [58] highlighted the uncertainty of indirect assessment to detect”

L563-564 rephrase “A similar applicability was also observed by this study when to estimate Ks for maize with CWSI.”

L566 “Bausch et al. [30] also obtained similar results”

L572-573 Can the authors expand on the following? “It is probable that even a well-watered crop could have a high canopy temperature because of other changes of microclimate, such as Ta and RH [29,34].”

L580 “Diversities, several studies [31,59,60]” It is not clear what the authors mean with “diversities”

L590-591 Rephrase “The difference of cumulative ET between model and observed were 2.6 mm, 8.9 mm, and 5 mm in three treatments, respectively.”

L591 “The result at long time scale confirmed” A long time scale study? Who did this study? Was it the previous work of the authors? From the UAV observations of the current study, it cannot be treated as a long time scale.

L598 Rephrase, “differently” is not the best word “Differently, UAV can monitor field ET information scale up information from the leaf to canopy/field levels and maybe a suitable technology for actual problems scouting within field scale.”

L600 “ maps (Figure 9), we can observe the evapotranspiration”

L605 Rates of what?

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-598317). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

 

Major comments:

 

Point 1: On line 471, the authors say that the remote sensing model was evaluated using ET-FAO data, does this mean that the evaluation was just a comparison to the tabular method? Or was there measured ET from an eddy covariance tower/lysimeter. If there were no actual observations, please include a comparison on the ET rates to similar studies.

 

Response 1: Thank you, unfortunately, there was no measured ET from an eddy covariance tower or lysimeter. In this manuscript, the remote sensing model was evaluated using the ET calculated by modified FAO-56 dual Kc method. Feng et al. [1] and Ding et al. [2] found good agreements between the predicted ET using the modified model and the measurements through lysimeter and covariance for maize and suggested the modified dual crop coefficient method could estimate maize ET accurately on the North China. Thus, employing the modified FAO-56 dual Kc method as the comparison data has certain accuracy. Reference [3,4] were also used the FAO-56 Kc method to validate the derived ET from their model.

And change has been made in 360-368.

 

Point 2: On line 596 “For instance, Landsat and SPOT have spatial resolution of 30 and 20 m, respectively.”  The authors only mention Landsat and SPOT, however, there are better satellites in use. An example is Sentinel 2 with a ~5 day retrieval period and 10 m resolution on the optical bands. Also, there are CubeSats with near daily acquisition time and 3 m spatial resolution. The authors should justify on the focus on Landsat and SPOT or include some references and discussion of new observation platforms.

 

Response 2: Thank you, change has been made in Line 603-607.

 

Point 3: On lines 598-599, the authors suggest that UAVs can monitor information up to the leaf scale. Is this feasible and economically viable for farmers? What would be the possible difficulties if so?

 

Response 3: Thank you, we have added the discussion about the UAVs’ feasible and economically viable for farmers in Line 623-627.

 

Point 4 : L599-602 “From the ET maps (Figure 9), we can observed the evapotranspiration of crops varied even with the same treatment. Table 4 showed the mean and CV values of different treatments due to the different soil texture and soil heterogeneity in the field.” The authors need to discuss the implications of this information rather than just repeating it from the results.

 

Response 4: Thank you, change has been made in Line 614-618.

 

 

Minor comments:

 

Point 1: What do the authors mean by “data conversion” on L126. Also, note that thermal imagery has the added burden of calibration which severely affects the absolute accuracy of the LST and can introduce other artifacts on the image.

 

Response 1: Thank you, change has been made in Line 126. And the inaccurate calibration of thermal imagery has been added in Line 544.

 

Point 2: On line 225 it was mentioned that a cubic spline was used to derive daily LAI and canopy height, can the authors explain why was a cubic spline used as the interpolation method? How can these parameters affect regions of high heterogeneity? Were they modified over those areas to be representative?

 

Response 2: Thank you, we collect the LAI and plant height of maize in the treatment zones and use the average value to represent the zones. But we didn’t measure these parameters every day. Because the original FAO-56 dual Kc model used linear interpolation to obtain daily Kcb and did not account into actual non-linear dynamic of Kcb due to variation of canopy cover, ET estimated by the original model may cause errors. Therefore, in order to evaluate the dynamic changes of ET in the maize field more accurately, the canopy height and LAI was used to modify the dynamic Kcb. Cubic spline interpolation is a piecewise regression approach that uses third-order polynomials for interpolation between a series of paired data points [5]. Reference [6] was also employ cubic spline to interpolate between harvest dates for estimation of daily LAI values. Thus, we obtained the daily through observed canopy height and LAI and cubic spline. And these parameters was use to run the modified FAO-56 dual Kc method.

And the reason why we employ the cubic spline has been added in Line 226-228.

 

Point 3: Please mention the method to correct reflectivity data from the grey plate on line 288.

 

Response 3: Thank you, change has been made in Line 294.

                                                                             

Point 4: On line 336, there is a mention to upper and lower canopy temperatures, how are these limits computed? Do they come from the sample ensemble? Are they measured at the same time? Or are they the minimum and maximum daily canopy temperatures?

 

Response 4: The lower limit, and upper limit of canopy–air temperature difference (Tc−Ta) came from the sample ensemble and measured at the same time. They could be calculated by the VPD (vapor pressure deficit) and vapor pressure gradient (VPG), respectively. The computing procedure and results can be found in reference [7]. Because the content is not the key points of this manuscript, we didn’t list it clearly.

 

Point 5: On lines 345 and 346, there seems to be two equations for the same variable, but the text only mentions the equation on line 346. Can the authors clarify which equation was used? The 0 and 1 values for CWSI seem to be flipped for the same intervals.

 

Response 5: Thank you, sorry for this poor typing mistake, we have corrected the right equation in Line 349 he equation (14) was used to estimated CWSI according to the reference [7], and the Equation (15) is from the Ks=1-CWSI [8].

 

Point 6: On lines 549 to 556, much of this information should not be part of the discussion. The authors should explain what are the effects of each method (direct or indirect) on the coefficients rather than provide an introduction to them. PET can be estimated, the difficulty is on measuring actual ET.

 

Response 6: Thank you, change has been made in Line 560-564.

 

Point 7: On lines 127 to 129, the authors state that “For example, due to the image resolution obtained by the two sensors is different, irrigation manager need to downscale the spatial resolution of multispectral map to match the scale of thermal imagery.” However, usually, the lower resolution sensor is the thermal one, furthermore, there is not a clear agreement on the required spatial resolution and how and when it begins to affect predictions based on UAV observations. And, most importantly, the biggest effect is not only on the pixel mixing due to the downscale, but on the georectification and coregistration of the images.

 

Response 7: Thank you, change has been made in Line 126 and 130. And we have added your comments in Line 128-130.

 

Point 8: L66 “However, the method is”.

 

Response 8: Thank you, change has been made in Line 66.

 

Point 9:  L68 due to inconsistent crop growth”.

 

Response 9: Thank you, change has been made in Line 67

 

Point 10: L72 “large irrigated fields”.

.

Response 10: Thank you, change has been made in Line 72.

 

Point 11: L82 Please add a few examples of these new technologies with their references.

 

Response 11: Thank you, change has been made in Line 82.

 

Point 12: L89 “when the crop completely covers”

 

Response 12: Thank you, change has been made in Line 88.

 

Point 13: L89 “Many studies have”

 

Response 13: Thank you, change has been made in Line 89.

 

Point 14: L101 “It is necessary to use a new indicator”.

 

Response 14: Thank you, change has been made in Line 101.

 

Point 15: L103 “canopy temperature is”.

 

Response 15: Thank you, change has been made in Line 103.

 

Point 16: L113 “that using an appropriate”.

 

Response 16: Thank you, change has been made in Line 114.

 

Point 17: L119-L120 Rephrase “which can be greatly constrained by labor intensive and time consuming”

 

Response 17: Thank you, change has been made in Line 120.

 

Point 18: L121 “estimate ET”.

 

Response 18: Thank you, change has been made in Line 122.

 

Point 19: L124 “CWSI, and water deficit index”.

 

Response 19: Thank you, change has been made in Line 124.

 

Point 20: L126-127 “obstacle to estimate ET with UAV systems”.

 

Response 20: Thank you, change has been made in Line 126.

 

Point 21: L129 “the Ks and Kcb may influence”

 

Response 21: Thank you, change has been made in Line 130.

 

Point 22: L135 Rephrase “using one multispectral UAV should be explored.” The UAV is not multispectral but the camera/sensor attached to it is.

 

Response 22: Thank you, change has been made in Line 136.

 

Point 23: L490-491,“This study aims to estimate more”.

 

Response 23: Thank you, change has been made in Line 137.

 

Point 24: L141-143 “through the relation between water stress index and VIs, to eliminate discrepancies caused by using data at different scales and obtain high resolution spatiotemporal ET maps.” The author should include the achieve resolution(s) in parenthesis, 10 cm for example.

 

Response 24: Thank you, which has been made in Line 144.

 

Point 25: L164 “MIK-2000H flow meters” provide the accuracy of the flow meters.

 

Response 25: Thank you, change has been made in Line 166-167.

 

Point 26: L214 Please specify that these ranges are for the thermal instrument.

 

Response 26: Thank you, change has been made in Line 216.

 

Point 27: L253 “the canopy height and LAI were used to modify the dynamic Kcb.” The use of canopy height rather than h is a personal preference as it makes the text more readable and is not an overly long word.

 

Response 27: Thank you, changes has been made in Line 222 and 228.

 

Point 28: L258 What is the value of k and where is it on equations 3-4?

 

Response 28: Thank you, the value of k is 0.7 (Table 2) and on the Equation (3). Changes has been made in Line 262.

 

Point 29: L324 “has been obtained through” Or add another method to obtain Kcb.

 

Response 29: Thank you, change has been made in Line 327.

 

Point 30: L327-329 “The two different Ks obtained from CWSI and Tc ratio were derived from the handheld infrared thermometer, with daily values taken between 11:00 and 13:00 (local time), which are approximate times of peak stress.”

 

Response 30: Thank you, change has been made in Line 330-331.

 

Point 31: L351 How is the variable Tc ratio estimated? What is this ratio?

 

Response 31: Thank you, change has been made in Line 350-358.

 

Point 32: L354 “have suggested that the modified dual”.

 

Response 32: Thank you, change has been made in Line 361.

 

Point 33: L357 “from the modified FAO-56”.

 

Response 33: Thank you, change has been made in Line 364.

 

Point 34: L390 “the highest Tc in TRT 2 reached to 34°C.”

 

Response 34: Thank you, change has been made in Line 401.

 

Point 35: Figure 4 Modify the x-title to include the year.

 

Response 35: Thank you, change has been made in Line 402.

 

Point 36: L445-446 Can you elaborate on this? “It’s probable that even a well-watered crop could have a high canopy temperature because of irrigation followed by very hot day.”

 

Response 36: Thank you, change has been made in Line 457. And we elaborate it in Line 579-582.

 

Point 37: L473-475 “It can be clearly seen from Figure 8 that the two different water stress indices model performed their evaluation results from the coefficient of determination (R2), root mean square error (RMSE) and index of agreement (d).” Rephrase, it is not clear what the authors meant with this sentence. Is it that the two stress indices were evaluated based on the previously mentioned statistics?

 

Response 37: Thank you, change has been made in Line 482-484.

 

Point 38: L476 “Figure 8 (a) and (b), is greater than 0.9, which”

 

Response 38: Thank you, change has been made in Line 484.

 

Point 39: L476 “ET based on water stress has a strong fit” A perfect fit would be for a value of one.

 

Response 39: Thank you, change has been made in Line 485.

 

Point 40: L478 Please include in parenthesis a NRMSE value which will give readers a better picture if this is a low or high value (normalize by the mean).

 

Response 40: Thank you, change has been made in Line 486 and 488.

 

Point 41: L500 “The different soil texture, and soil heterogeneity lead”.

 

Response 41: Thank you, change has been made in Line 509.

 

Point 42: L506 Rephrase “which illustrated that the prolonged water stress in anomaly distinctly amplified spatial fluctuations in field soil heterogeneity via its influences on maize evapotranspiration condition.”

 

Response 42: Thank you, change has been made in Line 515.

 

Point 43: L519 “were calculated by”

 

Response 43: Thank you, change has been made in Line 527.

 

Point 45: L521 “As mentioned above, crop coefficients was derived”.

 

Response 45: Thank you, Thank you, change has been made in Line 529.

 

Point 46: L524 “the total water consumption of the three treatments”.

 

Response 46: Thank you, change has been made in Line 532.

 

Point 47: L526 “ET calculated by the Vis”.

 

Response 47: Thank you, change has been made in Line 534.

 

Point 48: L527 Rephrase “Note that the retrieved ET were familiar with the crop water consumption.” It is not clear what the authors mean with “familiar”.

 

Response 48: Thank you, change has been made in Line 535.

 

Point 49: “of models for ET assessment.”.

 

Response 49: Thank you, change has been made in Line 541.

 

Point 50: L533-534 Rephrase “Models include surface energy balance model and strictly empirical VI model.”

 

Response 50: Thank you, change has been made in Line 541 and 542.

 

Point 51: L534 “Though the application of surface energy balance models are able to” Who is able? Who is the subject of this sentence?

 

Response 51: Thank you, change has been made in Line 543.

 

Point 52: L535 “accuracy. Deficiencies” This is a separate sentence.

 

Response 52: Thank you, change has been made in Line 543.

 

Point 53: L541-542 “Multispectral VIs calculated from canopy reflectance can be used to simulate real-time Kcb.”

 

Response 53: Thank you, change has been made in Line 550.

 

Point 54: L543 “help of UAVs”

 

Response 54: Thank you, change has been made in Line 551.

 

Point 55: L543 Which studies? Add the appropriate references.

 

Response 55: Thank you, change has been made in Line 552.

 

Point 56: L555-556 rephrase “In many situation, it's impossible to get the data of PET and actual ET when lack of.

 

Response 56: Thank you, change has been made in Line 560 and 561.

 

Point 57: L556 “Diarra et al. [58] highlighted the uncertainty of indirect assessment to detect”

 

Response 57: Thank you, change has been made in Line 562-563.

 

Point 58: L563-564 rephrase “A similar applicability was also observed by this study when to estimate Ks for maize with CWSI.”

 

Response 58: Thank you, change has been made in Line 569.

 

Point 59: L566 “Bausch et al. [30] also obtained similar results”

 

Response 59: Thank you, change has been made in Line 572.

 

Point 60: L572-573 Can the authors expand on the following? “It is probable that even a well-watered crop could have a high canopy temperature because of other changes of microclimate, such as Ta and RH [29,34].”

 

Response 60: Thank you, change has been made in Line 579-582.

 

Point 61: L580 “Diversities, several studies [31,59,60]” It is not clear what the authors mean with “diversities”.

 

Response 61: Thank you, change has been made in Line 589.

 

Point 62: L590-591 Rephrase “The difference of cumulative ET between model and observed were 2.6 mm, 8.9 mm, and 5 mm in three treatments, respectively.”

 

Response 62: Thank you, change has been made in Line 599 and 600.

 

Point 63: L591 “The result at long time scale confirmed” A long time scale study? Who did this study? Was it the previous work of the authors? From the UAV observations of the current study, it cannot be treated as a long time scale.

 

Response 63: Thank you, change has been made in Line 600.

 

Point 64: L600 “maps (Figure 9), we can observe the evapotranspiration”

 

Response 64: Thank you, change has been made in Line 609.

 

Point 65: L605 Rates of what?

 

Response 65: Thank you, which has been deleted.

 

 

We are very grateful to you for your valuable and elaborate comments about this manuscript, which is of great significance to our work. Thank you again for your help with this manuscript!

 

 

Reference

 

Feng, Y.; Cui, N.; Gong, D.; Wang, H.; Hao, W.; Mei, X. Estimating rainfed spring maize evapotranspiration using modified dual crop coefficient approach based on leaf area index. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2016, 32, 90-98, doi:10.11975/j.issn.1002-6819.2016.09.013. Ding, R.; Kang, S.; Zhang, Y.; Hao, X.; Tong, L.; Du, T. Partitioning evapotranspiration into soil evaporation and transpiration using a modified dual crop coefficient model in irrigated maize field with ground-mulching. Agricultural Water Management 2013, 127, 85-96, doi:https://doi.org/10.1016/j.agwat.2013.05.018. Elnmer, A.; Khadr, M.; Kanae, S.; Tawfik, A. Mapping daily and seasonally evapotranspiration using remote sensing techniques over the Nile delta. Agricultural Water Management 2019, 213, 682-692, doi:https://doi.org/10.1016/j.agwat.2018.11.009. French, N.A.; Hunsaker, J.D.; Bounoua, L.; Karnieli, A.; Luckett, E.W.; Strand, R. Remote Sensing of Evapotranspiration over the Central Arizona Irrigation and Drainage District, USA. Agronomy 2018, 8, doi:10.3390/agronomy8120278. Sadler, E.; Bauer, P.J.; Busscher, W.J.; Millen, J.A. Site-Specific Analysis of a Droughted Corn Crop. Agronomy Journal 2000, 92, doi:10.2134/agronj2000.923403x. Gramig, G.; Stoltenberg, D.; Norman, J. Weed species radiation-use efficiency as affected by competitive environment. Weed Science - WEED SCI 2006, 54, 1013-1024, doi:10.1614/WS-06-012R.1. Zhang, L.; Zhang, H.; Niu, Y.; Han, W. Mapping Maize Water Stress Based on UAV Multispectral Remote Sensing. Remote Sensing 2019, 11, doi:10.3390/rs11060605. Jackson, R.D.; Idso, S.B.R.J.; Reginato, R.J.; Pinter; P., J.J.W.R.R. Canopy Temperature as a Crop Water Stress Indicator. Water Resources Research 1981, 17, 1133-1138.

Author Response File: Author Response.docx

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.

Round 1

Reviewer 1 Report

The paper is sound and with good field experiments, but I do not see anything innovative worthy publishing. The only difference with previously similar papers is the type of crop, but the approach is almost the same. For example this paper is similar with the following three papers:

https://www.mdpi.com/2072-4292/11/6/605/htm

https://www.hydrol-earth-syst-sci.net/20/1523/2016/hess-20-1523-2016.pdf

https://www.researchgate.net/publication/331734201_Mapping_Maize_Water_Stress_Based_on_UAV_Multispectral_Remote_Sensing

https://www.researchgate.net/publication/320116049_UAV_AND_TIRS-LANDSAT-8_USING_FOR_ACTUAL_EVAPOTRANSPIRATION_ESTIMATION_ON_CHARDONNAY_VINEYARD_IN_PINTO_BANDEIRA_RS_BRAZIL

For this reason I would suggest that the paper be a research report and not a journal publication. Unless if the authors develop, maybe, a high resolution ET map from their research maybe. In my understanding this is non existing except the MOD16 at 250m resolution and GLEAM at 25K which are not suitable for small fields. Or else the authors should justify the difference of their study with previously published work.

The paper is badly structures. For example, from line 149 to 176 the authors write about how they conducted their experiment, but this is the introduction. This section should be part of the introduction.

The use of the English language is also very poor. It should be edited by a native English speaker

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-538843). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

Point 1: The paper is sound and with good field experiments, but I do not see anything innovative worthy publishing. The only difference with previously similar papers is the type of crop, but the approach is almost the same. Or else the authors should justify the difference of their study with previously published work. For example this paper is similar with the following three papers:

https://www.mdpi.com/2072-4292/11/6/605/htm

https://www.hydrol-earth-syst-sci.net/20/1523/2016/hess-20-1523-2016.pdf

https://www.researchgate.net/publication/331734201_Mapping_Maize_Water_Stress_Based_on_UAV_Multispectral_Remote_Sensing

https://www.researchgate.net/publication/320116049_UAV_AND_TIRS-LANDSAT-8_USING_FOR_ACTUAL_EVAPOTRANSPIRATION_ESTIMATION_ON_CHARDONNAY_VINEYARD_IN_PINTO_BANDEIRA_RS_BRAZIL

Response 1: Thank you, we have carefully read the articles you have recommended, and believe that these articles have great significance in the application of agricultural remote sensing. On the whole, our research method is different from these articles. Although the main target of our study and these articles are evapotranspiration estimation, we believe that our work is an important supplement to UAV-guided agricultural irrigation. Specific differences and reasons are as follows:

First, as you suggested in the first article “Mapping Maize Water Stress Based on UAV Multispectral Remote Sensing”, which was the previous research of our group. In this paper, we demonstrated the potentiality of using high-resolution UAV multispectral imagery to map maize water stress. However, water stress index can only be used as an indicator of crop water shortage and can’t determine how much water should be irrigated [1]. Based on which, in order to convert qualitative assessment of crop water stress to quantitative assessment of crop water demand. We done more in-depth research and studied the possibility using multispectral UAV to monitor crop water requirement.

Secondly, we also noticed that the methods used in the rest articles to estimate crop evapotranspiration were different from this paper, they used energy balance equations while we employed the FAO crop coefficients method. The energy balance equation does have a strong physical foundation and can obtain accurate ET values. But it requires a lot of complex parameters and multiple (thermal and multispectral) sensors, and the temperature data are vulnerable to weather [2]. Meanwhile, thermal infrared UAV system that have the problems of unstable temperature acquisition, low mosaic precision and needing complex pretreatments (radiometric calibration, temperature correction and canopy temperature extraction) [3]. Compared to the thermal infrared system, a multispectral remote-sensing system has the advantages of stable information acquisition and mature mosaic technology [4]. In addition, the FAO approach for estimating crop evapotranspiration requires fewer inputs and theoretical background knowledge, thus it is simpler than the energy balance [5]. FAO crop coefficients method has been widely used in evapotranspiration estimation [6]. Research on crop evapotranspiration based on UAV multi-spectral remote sensing is rare and has strong applicability and reliability.

Thus, we believe that high-precision resolution, simpler and more suitable methods for popularization will contribute towards the increasing interest in the use of UAVs for irrigation and crop management and are of great significance to the development of precision agriculture.

 

Finally, Discussions about energy balance equations and FAO crop coefficient, as well as the innovations in this article, were added to “4 Discussion” (Lines 501-516).

Point 2: For this reason I would suggest that the paper be a research report and not a journal publication. Unless if the authors develop, maybe, a high resolution ET map from their research maybe. In my understanding this is non existing except the MOD16 at 250m resolution and GLEAM at 25K which are not suitable for small fields.

Response 2: Thank you, we illustrated that the spatial resolution of our multispectral camera is 4.7 cm (line 301), and obtained high resolution ET map (Figure 7). We believe that this resolution is suitable for farm-scale crop agricultural water management. The satellite data proposed by reviewer, such as Landsat (15m) and MODIS (250m), are more suitable for large-scale crop monitoring. The merits and demerits of satellite data have been discussed in the introduction (Lines 131-134).  

Point 3: The paper is badly structures. For example, from line 149 to 176 the authors write about how they conducted their experiment, but this is the introduction. This section should be part of the introduction.

Response 3: Thank you, changes were made.

Point 4：The use of the English language is also very poor. It should be edited by a native English speaker

Response 4: English language edited by MDPI Editing Service.

 

Reference

1.             Bausch, W.; Trout, T.; Buchleiter, G. Evapotranspiration adjustments for deficit-irrigated corn using canopy temperature: A concept. Irrigation and Drainage 2011, 60, 682-693, doi:10.1002/ird.601.

2.             Xia, T.; Kustas, W.P.; Anderson, M.C.; Alfieri, J.G.; Gao, F.; McKee, L.; Prueger, J.H.; Geli, H.M.E.; Neale, C.M.U.; Sanchez, L., et al. Mapping evapotranspiration with high-resolution aircraft imagery over vineyards using one- and two-source modeling schemes. Hydrol. Earth Syst. Sci. 2016, 20, 1523-1545, doi:10.5194/hess-20-1523-2016.

3.             Ribeiro-Gomes, K.; Hernández-López, D.; Ortega, J.F.; Ballesteros, R.; Poblete, T.; Moreno, M.A. Uncooled Thermal Camera Calibration and Optimization of the Photogrammetry Process for UAV Applications in Agriculture. Sensors (Basel) 2017, 17, 2173, doi:10.3390/s17102173.

4.             Shi, X.; Han, W.; Zhao, T.; Tang, J. Decision Support System for Variable Rate Irrigation Based on UAV Multispectral Remote Sensing. Sensors 2019, 19, doi:10.3390/s19132880.

5.             González-Dugo, M.; Mateos, L. Spectral Vegetation Indices For Estimating Cotton And Sugarbeet Evapotranspiration; 2006; Vol. 852, pp. 115-123.

6.             Gontia, N.K.; Tiwari, K.N. Estimation of Crop Coefficient and Evapotranspiration of Wheat (Triticum aestivum) in an Irrigation Command Using Remote Sensing and GIS. Water Resources Management 2009, 24, 1399-1414, doi:10.1007/s11269-009-9505-3.

 

Author Response File: Author Response.docx

Reviewer 2 Report

The study investigated the performance of crop stress factors generated from remotely sensed data (UAV) for irrigation water management applications on a center pivot field, Northern China Plain. The study concluded the feasibility of using UAV for farm level monitoring and prescription of deficit irrigation management.

 

The motivation and experiment setup is well thought out and described well with useful practical applications in water management and precision agriculture.  I believe this will contribute towards the increasing interest in the use of UAVs for irrigation and crop management.

 

The manuscript will benefit from a more detailed description on some aspects of the methods. For example, it is not clear how Equation 5 is implemented.  Mainly, how “Dr” is determined. Is it from daily water balance model etc?  The same with the determination of  dTul and DTll. These are critical parameters and the reference cited [38] does not seem very explicit about it. 

 

Another concern is the lack of independent validation from, say, eddy covariance or lysimeter data.  The soil moisture collected at different depths was not interpreted/discussed as much. The water balance was supposed to serve this purpose, but its implementation begs the question why you did not at least do a seasonal total (after daily interpolation and/or modeling) rather than use 3 to 5 days of ET. The fact that the points in figure 6 represent different aggregation periods bring confusion with the time scale of comparison and potentially pooling the data at different time scales may inflate the correlation, i.e., 3 days of aggregation will plot lower and 5-day aggregation will plot higher. The integration of the UAV data to improve daily soil water simulation would have been a great use of such UAV data. Not sure if this was considered outside of the scope of the study.

 

Scaling and operation:

While the promise of UAV is great in terms of high spatiotemporal resolution, it will be useful to also discuss the challenges associated with scaling to larger areas and also the management aspect of handling the volume of such high frequency data for operational applications.

Finally, it is still not clear how the such data is used for irrigation scheduling. There was some discussion on the importance of converting the stress index into soil water amounts, but more discussion will be useful how this information can be converted into decision making information for irrigation scheduling.

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-538843). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

 

Point 1: For example, it is not clear how Equation 5 is implemented. Mainly, how “Dr” is determined. Is it from daily water balance model etc?

 

Response 1: Thank you,, and we used the Equation (R2) FAO-33 equivalent expression to Equation (R1) to calculate Ks. Daily soil moisture can be obtained by spline interpolating through the measured soil moisture data.

                         (R1)

 (R2)  

 (3)

: The threshold soil water, : soil water content

 

Point 2: The same with the determination of dTul and DTll. These are critical parameters and the reference cited [38] does not seem very explicit about it.

 

Response 2: Thank you, the calculation procedures and results of these two parameters are from [29], which was also the work of our research group. I have revised the reference in the paper.

 

Point 3: Another concern is the lack of independent validation from, say, eddy covariance or lysimeter data.

 

Response 3: Thank you, there are many articles using eddy covariance or lysimeter, to obtain accurate daily ET values. However, their popularization and application are limited by the complex design technology and high cost of various sensors. Eddy covariance are mostly used to detect large areas of crops and require uniform underlying surfaces [7]. But our sample areas are small and close with each other. Meanwhile, since our experiment is divided into three irrigation levels, it is difficult to monitor simultaneously with lysimeter. We also need to collect several data in the sample area. Artificial interference on the lysimeter will destroy the soil structure and affecting the accuracy of the data. Calculating ET based on traditional gravimetric method and water balance is the most traditional, effective method though it is time-consuming and laborious. Soil drying method is the basis of other soil moisture determination methods and often used to calibrate soil moisture sensor data, which is simple and accurate [8]. Thus, data from gravimetric method and water balance has certain validation value.

 

Point 4: The soil moisture collected at different depths was not interpreted/discussed as much. The water balance was supposed to serve this purpose, but its implementation begs the question why you did not at least do a seasonal total (after daily interpolation and/or modeling) rather than use 3 to 5 days of ET. The fact that the points in figure 6 represent different aggregation periods bring confusion with the time scale of comparison and potentially pooling the data at different time scales may inflate the correlation, i.e., 3 days of aggregation will plot lower and 5-day aggregation will plot higher.

 

Response 4:

Thank you,

(1)    We have added more information about soil moisture in 2.2.1 (Lines: 231-236).

 

(2) We measured soil water content two or three times each week for cumulated evapotranspiration calculation, and we used these intervals for comparison. Because of the continuous drought weather, the irrigation frequency during our experiment was relatively high, so we can only collect the interval data of 3-5 days and be unable to interpolate long time series. If interpolated, the results showed that the water consumption of these small periods is the same every day. When validated with these data, a large amount of data will focus on a vertical line in the image. However, there are relatively high correlation through soil water balance calculations for intervals during two successive measurement. Thus, we considered this way may more realistic and meets our experimental requirements and have certain significance in guiding irrigation.

 

(3) Although there is no daily data, we can determine the irrigation based on the amount of water consumed during this period. In many practical irrigation management, supplementary irrigation is based on the consumption of the previous period (e.g. 5 days). Thus, cumulative evapotranspiration comparison over intervals can still serve irrigation management decision.

 

Point 5: The integration of the UAV data to improve daily soil water simulation would have been a great use of such UAV data. Not sure if this was considered outside of the scope of the study.

 

Response 5: Thank you, as you said, soil moisture is a key component of soil water balance, and there are many researchers have generated soil moisture estimates using UAV. At present, microwave remote sensing has high reliability for monitoring soil moisture content. However, limited by factors of economic and technological. Most of researchers studied surface soil moisture through optical sensor placed on-board aircrafts [9], as well as, our team has previously studied the relationship between soil moisture content and vegetation index at 0-60 cm depth, and achieved good results. As for crop evapotranspiration, we need to consider root water uptake, and the depth of soil tillage layer varies with the growth period. According to the FAO-56 the maximum root depth of maize can reach to 80-3000px. On the other hand, which may be helpful for soil evaporation calculation. As the important parameter for farm management, this proposal may provide a new solution for our next step of soil evaporation estimation.

 

Point 6: While the promise of UAV is great in terms of high spatiotemporal resolution, it will be useful to also discuss the challenges associated with scaling to larger areas and also the management aspect of handling the volume of such high frequency data for operational applications.

 

Response 6: Thank you. The challenges of UAV remote sensing was added in “4 Discussion” (Lines 492-500).

 

Point 7: Finally, it is still not clear how the such data is used for irrigation scheduling. There was some discussion on the importance of converting the stress index into soil water amounts, but more discussion will be useful how this information can be converted into decision making information for irrigation scheduling.

 

Response 7: Thank you, as you said, how to apply our results to practical irrigation applications is very important. Our team has applied this result to the practical variable rate irrigation system based on UAV multispectral remote sensing with the following hierarchical framework (Figure R1). We input ETc, CWSI and precipitation into this variable irrigation systems and irrigate according to real-time crop conditions. And We added this in “4 Discussion”(Lines 487-490). More specific information can be found in paper [4]. This system employ the single crop coefficient, while we further consider the influence of water stress coefficient (Ks). We believe that this is a good improvement to the decision-making system.

Figure R1. Schematic representation of the decision support system for variable rate irrigation (DSS-VRI).

The DSS-VRI operational procedures include four parts. Their functions are as follows. Part 1 is to provide unmanned aerial vehicle (UAV) multispectral image and meteorological data as input. Part 2 processes and selects data from Part 1, figures out the crop evapotranspiration model (ETc), crop water stress index (CWSI) and precipitation. The data input to the fuzzy system. Part 3 shows the work flow of the fuzzy system. Part 4 depicts the duty-cycle control map for a partial management zone.

 

 

Reference

1.             Culf, A.D.; Foken, T.; Gash, J.H.C. The Energy Balance Closure Problem; 2004.

2.             Shukla, A.; Panchal, H.; Mishra, M.; Patel, P.; Srivastava, H.; Patel, P.; Shukla, A.K. Soil Moisture Estimation using Gravimetric Technique and FDR Probe Technique: A Comparative Analysis; 2014; Vol. 14, pp. 89-92.

3.             Hassan-Esfahani, L.; Torres-Rua, A.; Ticlavilca, A.M.; Jensen, A.; McKee, M. Topsoil moisture estimation for precision agriculture using unmmaned aerial vehicle multispectral imagery. In Proceedings of Geoscience & Remote Sensing Symposium.

4.             Shi, X.; Han, W.; Zhao, T.; Tang, J. Decision Support System for Variable Rate Irrigation Based on UAV Multispectral Remote Sensing. Sensors 2019, 19, doi:10.3390/s19132880.

 

Author Response File: Author Response.docx

Reviewer 3 Report

No comments.

Author Response

Dear Editors and Reviewers:

 

Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “UAV Multispectral Imagery Combined with FAO-56 Dual Approach for Maize Evapotranspiration Mapping in the North China Plain” (remotesensing-538843). These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significances to our researches. We have studied comments carefully and have made revision using the "Track Changes" function in Microsoft Word. The main corrections in the paper and the responds to the reviewer’s comments are as flows:

 

 

Thank you, manuscript was majorly revise according to comments of editors and reviewers, and English language edited by MDPI Editing Service.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The manuscript has been improved and the authors attended to all my previous comments. Best of luck

Reviewer 3 Report

-