Evaluation of Multiple Methods for the Production of Continuous Evapotranspiration Estimates from TIR Remote Sensing
Round 1
Reviewer 1 Report
It is a good work presented , analyses different ways to interpolate missing ET from Satelite data in order to improve the water needs estimation in crops.
Despite the results in BIAS are important authors should present also another statistical parameter to show off the variability of the results (Root Mean Square Error or a simple standar deviation.
- Improve the quality image of Figure 3. May the images copied in low quality?
- Re-think o re-build Figure 5. Authors should change vertical axes to improve the reading of the data presented. Is Kairouan 2012 obtained the better results to show?
- Authors should specify the units of the X-axis Figure 5 and 6
Author Response
Dear Editors, dear colleagues,
We first want to thank you for reviewing our paper and we are pleased to submit our revised manuscript.
Minor corrections were included in this revised version and our responses are addressed below.
Reviewer 3 / Comment #1: Despite the results in BIAS are important authors should present also another statistical parameter to show off the variability of the results (Root Mean Square Error or a simple standar deviation.
We made the choice to present RMSE only in table 3 to show the error in the reconstruction of daily ET from an instantaneous measurement because, at seasonal scale, the metric is linked to the cumulative seasonal ET. Indeed, we found complicated to understand the square root of the average of squared errors which will smooth outliers for a cumulative quantity.
Reviewer 3/ Comment #2: Improve the quality image of Figure 3. May the images copied in low quality?
Figure 3 replaced with a better quality.
Reviewer 3/ Comment #3: Re-think o re-build Figure 5. Authors should change vertical axes to improve the reading of the data presented. Is Kairouan 2012 obtained the better results to show? Authors should specify the units of the X-axis Figure 5 and 6
Done. Kairouan 2012 was chosen to illustrate the underestimation done on cloudy days and not especially for the quality of the result on this particular site.
Sincerely yours,
On behalf of all co-authors,
Emilie Delogu
Reviewer 2 Report
This is an interesting paper that addresses important questions about how to use temporally low-density remote sensing data to estimate daily and seasonal evapotranspiration. It has important findings about the relative merits of different interpolation supporting variables and the effect of image frequency on the accuracy of ET estimation. The paper also serves as a guide to simple interpolation models for ET estimation. The paper is acceptable with editorial revisions. However, there are several editorial issues with the paper
The terminology of “supports” to indicate interpolation supporting variables is odd and not very descriptive. I don’t think I have seen this word used in this way before. I would suggest substituting something like “interpolation supporting variables” and then give them the acronym ISVs throughout the rest of the paper.
It isn’t clear how the “supports” were chosen. This needs to be explained. What criteria were used to select them? Why were they all analyzed individually, rather than considering multivariate supporting models? Why wasn’t soil moisture included among the supporting variables?
Remember that all figures should be able to stand alone with their caption. Figure 3 doesn’t stand alone. It can’t be interpreted without going back to the text.
Citations are provided using both numbers (per journal specifications), but many citations are provided using author and year. Furthermore, some citations are missing in the literature cited section. This all needs to be cleaned up.
The first two sentences of the results section are journal guidelines that were not removed from the text.
Most of the paper is well-written, but it still needs careful editing throughout.
Author Response
Dear Editors, dear colleagues,
We first want to thank you for reviewing our paper and we are pleased to submit our revised manuscript.
Minor corrections were included in this revised version and our responses are addressed below.
Reviewer 1 / Comment #1: The terminology of “supports” to indicate interpolation supporting variables is odd and not very descriptive. I don’t think I have seen this word used in this way before. I would suggest substituting something like “interpolation supporting variables” and then give them the acronym ISVs throughout the rest of the paper.
We change the terminology and replace it with “reference quantity” (q) as Alfieri et al. (2017) named it in their paper dealing with the same topic as ours.
Reviewer 1 / Comment #2: It isn’t clear how the “supports” were chosen. This needs to be explained. What criteria were used to select them? Why were they all analyzed individually, rather than considering multivariate supporting models? Why wasn’t soil moisture included among the supporting variables?
The reference quantities were chosen to pursue the study of Alfieri et al. (2017). We then choose the same classical reference quantities as Alfieri et al. (2017) and we decided to add two new ones such as the combination of one classical with a proxy derived from the amount of precipitation in order to constrain the surface water status at the daily time scale. A simple, available and easily measurable reference quantity that could be introduced into the temporal upscaling to better take the surface water status into account is the precipitation information. (l 138 to 150)
We choose precipitation rather than soil moisture because precipitation is a simple available and easily measurable quantity while soil moisture measurement is more challenging.
Reviewer 1 / Comment #3: Figure 3 doesn’t stand alone. It can’t be interpreted without going back to the text.
We change the caption for a more relevant one.
Reviewer 1 / Comment #4 and #5: Citations are provided using both numbers (per journal specifications), but many citations are provided using author and year. Furthermore, some citations are missing in the literature cited section. This all needs to be cleaned up.
The first two sentences of the results section are journal guidelines that were not removed from the text.
Cleaned up !
Sincerely yours,
On behalf of all co-authors,
Emilie Delogu
Reviewer 3 Report
- This study compared the performance of some seasonal ET estimation models based on satellite images, and discussed the importance of interpolation estimation technology in the estimation error of ET. The contents involved remote sensing, hydrology and environmental monitoring, which is important and can arouse readers’ interests.
- The model estimation methods mainly integrate band characteristics of satellite images, energy balance model, temporal and spatial correlation between environmental factors, extrapolation and interpolation algorithms, and has complexity and high academic research value in process construction. In terms of research results, there is no question of arbitrarily determining whether a particular model is superior or inferior because it only conducts comparative analysis for specific regions. However, it is still recommended to discuss the feasibility of applying this model to other regions.
- This ET estimation methods still relied heavily on artificially set models, but both the satellite image contents and the model itself have great uncertainties, so in addition to boxplots showing the range distribution of the inferred results, it is recommended to import Montenegro Carlo analysis or probability models to discuss whether the research results are concentrated and convergent.
- The spatial resolution of satellite imagery is low, so to some extent its application results have the characteristics of homogenization, and the calculation process is mainly based on the linear interpolation model, whether it will affect the extreme value of ET is worth discussing.
5. The chart drawing quality is poor, it is recommended to increase the resolution, and it is recommended to place satellite images and estimated results images in the manuscript, so that readers can understand the conversion process more clearly. - The discussion process of this type of model should focus on the applicability analysis of "radiation-based" and "temperature-based", because the models developed by the two theoretical foundations in different climate regions may have great differences in results.
- The influence of terrain on the research results has not yet been discussed.
Author Response
Dear Editors, dear colleagues,
We first want to thank you for reviewing our paper and we are pleased to submit our revised manuscript.
Minor corrections were included in this revised version and our responses are addressed below.
Reviewer 2 / Comment #1: The model estimation methods mainly integrate band characteristics of satellite images, energy balance model, temporal and spatial correlation between environmental factors, extrapolation and interpolation algorithms, and has complexity and high academic research value in process construction. In terms of research results, there is no question of arbitrarily determining whether a particular model is superior or inferior because it only conducts comparative analysis for specific regions. However, it is still recommended to discuss the feasibility of applying this model to other regions.
The feasibility of applying this model to other region is actually the aim of another study conducted at CESBIO as this paper underlined that the choice of the reference quantity is dependent on the objective (l.538). The methods developed should indeed be tested and transferred to other ecosystems and focused on product development at the spatial (plot) and temporal (daily) scales adapted for monitoring the water stress.
Reviewer 2 / Comment #2: This ET estimation methods still relied heavily on artificially set models, but both the satellite image contents and the model itself have great uncertainties, so in addition to boxplots showing the range distribution of the inferred results, it is recommended to import Montenegro Carlo analysis or probability models to discuss whether the research results are concentrated and convergent.
The models used are not calibrated over the sites and we made the choice to study generic and very simple models which can theoretically applied anywhere operationally. We did not use satellite images but instantaneous estimates of ET (model SPARSE) from a surface temperature measured in situ.
Reviewer 2 / Comment #3: The spatial resolution of satellite imagery is low, so to some extent its application results have the characteristics of homogenization, and the calculation process is mainly based on the linear interpolation model, whether it will affect the extreme value of ET is worth discussing.
The chart drawing quality is poor, it is recommended to increase the resolution, and it is recommended to place satellite images and estimated results images in the manuscript, so that readers can understand the conversion process more clearly.
We did not use satellite imagery, only in situ data, I guess there is no issue of spatial homogenization but I am not sure I understand the comment.
Reviewer 2 / Comment #4: The discussion process of this type of model should focus on the applicability analysis of "radiation-based" and "temperature-based", because the models developed by the two theoretical foundations in different climate regions may have great differences in results.
There was significant variability in the performances from site to site particularly for long revisit frequencies (over 8 days). The specific causes of these differences are not fully understood, and we try but can’t find an applicability analysis which could explain these differences. (Page 16 )
Reviewer 2 / Comment #5: The influence of terrain on the research results has not yet been discussed.
Dry regions (as Kairouan and Haouz sites) with low ET fluxes seemed to be less affected by the degradation of the revisit frequency than other regions where ET is higher (Tables 4 and 5), which is quite logical when considering that the variability of ET are lower (Page 16). For any other sites, the influence of terrain did not appear preponderant in the quality of the results which was strongly and mainly affected by the revisit frequency more than the climatic area.
Sincerely yours,
On behalf of all co-authors,
Emilie Delogu
