Sensitivity of Surface Fluxes in the ECMWF Land Surface Model to the Remotely Sensed Leaf Area Index and Root Distribution: Evaluation with Tower Flux Data

Round 1

Reviewer 1 Report

Review on « High sensitivity of surface fluxes to root distribution in the CHTESSEL model » by Stevens et al.

The manuscript describes a sensitivity study of sensible and latent heat fluxes predicted by CHTESSEL to LAI and some vegetation parameters including minimum resistance in the Jarvis formulation and root distribution and depth. The objective is to evaluate the potential of improvement of fluxes prediction, in particular during hydric stress periods. This is an important subject both for NWP in arid and semi-arid regions and for land surface modeling in a climate change context. The main results is the strong sensitivity of surface fluxes to root distribution. The exponential distribution concentrating roots in the upper layers is usually used in the land surface models. When the incoming radiation is high, the upper soil layers dry out quickly and evapotranspiration is stopped. By distributing uniformly the roots within the profile, the root-zone reservoir is indirectly increased and the evapotranspiration stop by lack of water is delayed by several days after a rainfall or an irrigation events. This is not a breaking news and the highlighted improvements are quite low and site-dependent but the paper is interesting, well written and well organized. It deserves publication for TESSEL users and will benefit to a broader audience including the SVAT community.

I have several comments listed below but as none of them are major, I recommend publication after minor revisions.

Minor comments

• Although the high sensitivity of surface fluxes to root distribution is the key result of this study, the topics adressed in the manuscript are well beyond this specific point. I would suggest to change the title to reflect better the content of the work.
• Please state clearly what are the objective of this work in the introduction
• Please report the depth of the four layers in the legend of table S1 to ease reading.
• At the first cross-reference to figure in the supplementary material (p. 6 « see Figure S14 »), please specify « in the supplementary material » to ease reading
• Equation 3 : check the third conditions Theta > (instead of <) ThetaCap ?
• 4 availability of MODIS LAI : 2002 or 2003 ? as there is no MODIS time series data in none of the figures S1-S17 for 2002.
• 7 « It is somehow similar to an estimate of potential evaporation » Not really as the stress to water vapor deficit is still taken into account and because if there is no water anymore in the root-zone, evapotranspiration will be stopped as well (Amplero in 2003 for instance).­­
• 7 model evaluation. The choice of the ranking approach is well argued and I agree with the authors that there is no need to implement a real calibration/validation approach to show the sensitivity of the model to Rsmin and rooting depth. Nevertheless, it would have been interesting to provide some quantitative elements on the differences in terms of parameter values between the 10 best-ranked simulations for instance. To strengthen the discussion, the authors could add some sentences to explain if the parameter values are close and if not, if there is any equifinality issues between the two parameters.
• 8 the comparison of the new LAI and albedo products derived from remote sensing with those of the control run CTL (section 3.1) is a bit confusing as several products and averaging windows are compared but not use for fluxes evaluation. There is also no clear conclusion from this inter-comparison. In the main text, I would suggest to focus on the comparison between CTL and high resolution MODIS LAI and albedo (the one that will be used for evaluating the impact on the predicted fluxes in the rest of the manuscript) and either discard or move to an appendix the « high resolution » versus « bounding box » discussion. This would ease the reading.
• 8 : start of section 3.2 « The use of the high resolution MODIS albedo had a negligible impact on model performance » add « not shown » at the end of the sentence.
• 10, 2^nd paragraph : « with » not « wise » ?
• I was also wondering if there wasn’t any shallow water table in none of the selected 17 sites  that could explain the discrepancies between observed and predicted latent heat fluxes during drough periods. Could you comment on this ?

Author Response

Reply to the review of “High sensitivity of surface fluxes to root distribution in the CHTESSEL mode”  atmosphere-1016709.

The authors would like to thank the two anonymous reviewers for their constructive comments and suggestions to the manuscript. Below the reviewer comments are in bold and the authors reply in normal text. 

Reviewer #1

The manuscript describes a sensitivity study of sensible and latent heat fluxes predicted by CHTESSEL to LAI and some vegetation parameters including minimum resistance in the Jarvis formulation and root distribution and depth. The objective is to evaluate the potential of improvement of fluxes prediction, in particular during hydric stress periods. This is an important subject both for NWP in arid and semi-arid regions and for land surface modeling in a climate change context. The main results is the strong sensitivity of surface fluxes to root distribution. The exponential distribution concentrating roots in the upper layers is usually used in the land surface models. When the incoming radiation is high, the upper soil layers dry out quickly and evapotranspiration is stopped. By distributing uniformly the roots within the profile, the root-zone reservoir is indirectly increased and the evapotranspiration stop by lack of water is delayed by several days after a rainfall or an irrigation events. This is not a breaking news and the highlighted improvements are quite low and site-dependent but the paper is interesting, well written and well organized. It deserves publication for TESSEL users and will benefit to a broader audience including the SVAT community.

I have several comments listed below but as none of them are major, I recommend publication after minor revisions.

We thank the reviewer for the positive and constructive review. A detailed reply to each comments 

Minor comments

• Although the high sensitivity of surface fluxes to root distribution is the key result of this study, the topics adressed in the manuscript are well beyond this specific point. I would suggest to change the title to reflect better the content of the work.

• Following the reviewer suggestion, as well as reviewer #2, the title was changed to “Sensitivity of surface fluxes in the ECMWF land surface model to remote sensed Leaf Area Index and roots distribution: evaluation with tower flux data. “

• Please state clearly what are the objective of this work in the introduction

• On the last paragraph of the introduction we identify the main objective of the study and the two points addressed: “The main objective of the study is to evaluate the representation of canopy resistance by testing (i) the added value of the high-resolution LAI and albedo to represent local station conditions and (ii) the impact of roots distribution on soil moisture stress.”

• Please report the depth of the four layers in the legend of table S1 to ease reading.

• Added the depth of the four layers to the legend of table S1.

• At the first cross-reference to figure in the supplementary material (p. 6 « see Figure S14 »), please specify « in the supplementary material » to ease reading

• Added 

• Equation 3 : check the third conditions Theta > (instead of <) ThetaCap ?

• Corrected

• 4 availability of MODIS LAI : 2002 or 2003 ? as there is no MODIS time series data in none of the figures S1-S17 for 2002.

• The MODIS LAI data is available since July 2002, but the climatologies were computed from 2003 onwards to guarantee complete years (clarified in section 2.1.2). Therefore, in the supplementary figures the “MODIS time serie” only start in 2003. 

• 7 « It is somehow similar to an estimate of potential evaporation » Not really as the stress to water vapor deficit is still taken into account and because if there is no water anymore in the root-zone, evapotranspiration will be stopped as well (Amplero in 2003 for instance).­­

• We thank the reviewer for pointing out this. We had to keep the limitation of water availability to avoid numerical instabilities in the model. To avoid confusion, this sentence was removed. 

• 7 model evaluation. The choice of the ranking approach is well argued and I agree with the authors that there is no need to implement a real calibration/validation approach to show the sensitivity of the model to Rsmin and rooting depth. Nevertheless, it would have been interesting to provide some quantitative elements on the differences in terms of parameter values between the 10 best-ranked simulations for instance. To strengthen the discussion, the authors could add some sentences to explain if the parameter values are close and if not, if there is any equifinality issues between the two parameters.

• Sensitivity to Rsmin and Rooting depth was performed independently, therefore we cannot discuss equifinality. The joint sensitivity of both parameters would allow such a discussion, but we avoided that on purpose to leave it clear that our intention was not to calibrate the model.  

• 8 the comparison of the new LAI and albedo products derived from remote sensing with those of the control run CTL (section 3.1) is a bit confusing as several products and averaging windows are compared but not use for fluxes evaluation. There is also no clear conclusion from this inter-comparison. In the main text, I would suggest to focus on the comparison between CTL and high resolution MODIS LAI and albedo (the one that will be used for evaluating the impact on the predicted fluxes in the rest of the manuscript) and either discard or move to an appendix the « high resolution » versus « bounding box » discussion. This would ease the reading.

• Following the reviewer suggestion, the “bounding box” was discarded from the evaluation of Albedo and LAI. The text was updated as well as tables 3 and 4. 

• 8 : start of section 3.2 « The use of the high resolution MODIS albedo had a negligible impact on model performance » add « not shown » at the end of the sentence.

• Added

• 10, 2nd paragraph : « with » not « wise » ?

• Sentence was confusing. We changed to: “It was not possible to identify a systematic improvement in model performance associated with the use of the high resolution LAI dataset.”

• I was also wondering if there wasn’t any shallow water table in none of the selected 17 sites  that could explain the discrepancies between observed and predicted latent heat fluxes during drough periods. Could you comment on this ?

• From the initial 20 sites used in Best et al 2015 the station Merbleue was excluded in this study as it is located in a wetland. This was causing large discrepancies between observed and modeled fluxes (explained in section 2.1.1). From the remaining 17 sites, we are not aware of further sites that have a shallow water table that could impact the fluxes. However, this could explain some of the discrepancies. The following was added in the discussion: “The presence ofa shallow water table could also influence latent heat flux during drought periods [61]. Water table isnot represented in CHTESSEL, and we are not aware of such effects in the 17 sites considered” 
• [61] Pinto, Clara A and Nadezhdina, Nadezhda and David, Jorge S and Kurz-Besson, Cathy and Caldeira,Maria C and Henriques, Manuel O and Monteiro, Fernando G and Pereira, João S and David, Teresa S,Transpiration in Quercus suber trees under shallow water table conditions: the role of soil and groundwater.Hydrological Processes2014,28, 6067–6079.  doi:10.1002/HYP.10097

Reviewer 2 Report

Review for atmosphere-1016709:  “High sensitivity of surface fluxes to root distribution in the CHTESSEL model” By Stevens et al.

This paper investigates the root distribution impacts on surface fluxes by conducting a set of sensitivity experiments using the ECWMF land surface model. The research is of interest to the research community considering the importance of land-atmosphere exchange to climate and weather. The manuscript is well-structured and clear-written. I have a few minor comments below. Since the journal technician or editor did not remind the authors to add line numbers to the manuscript, I tried my best to add the comments and hope they are clear to the authors.

• Title: I would suggest the authors change the word ‘CHTESSEL’ to a more general phase to increase visibility, considering ‘CHTESSEL’ does not sound familiar to most of the readers. (ECWMF land surface model?)
• The second line in the abstract: ‘in the Earth System. The’----extra space
• Last line in the abstract: ‘other LSM’----‘other LSMs’
• Last Paragraph in Section 1: ’17 FLUXNET’----Add one sentence to briefly introduce FLUXNET
• Section 2.1.2: How Quality Control (QC) is defined and calculated? How to interpret the values of QC?
• Section 2.2: ‘the canopy resistance (r_c) [23]’----‘the canopy resistance (r_c) [23] is calculated as’
• The table caption for table 2 is not correct.
• Table 3 and Table 4: Is there a way to transform the tables into figures? Figures should be more visually clear to the readers instead of stacks of numbers.

Author Response

Reply to the review of “High sensitivity of surface fluxes to root distribution in the CHTESSEL mode”  atmosphere-1016709.

The authors would like to thank the two anonymous reviewers for their constructive comments and suggestions to the manuscript. Below the reviewer comments are in bold and the authors reply in normal text. 

This paper investigates the root distribution impacts on surface fluxes by conducting a set of sensitivity experiments using the ECWMF land surface model. The research is of interest to the research community considering the importance of land-atmosphere exchange to climate and weather. The manuscript is well-structured and clear-written. I have a few minor comments below. Since the journal technician or editor did not remind the authors to add line numbers to the manuscript, I tried my best to add the comments and hope they are clear to the authors.

We thank the reviewer for the positive comments. We apologize for the missing line numbers. It was in the original latex version submitted. 

• Title: I would suggest the authors change the word ‘CHTESSEL’ to a more general phase to increase visibility, considering ‘CHTESSEL’ does not sound familiar to most of the readers. (ECWMF land surface model?)

• Following also reviewer #1 comment, the title was changed to: “Sensitivity of surface fluxes in the ECMWF land surface model to remote sensed Leaf Area Index and roots distribution: evaluation with tower flux data. “ 

• The second line in the abstract: ‘in the Earth System. The’----extra space

• This is an automatic space added by latex to keep the text justified. 

• Last line in the abstract: ‘other LSM’----‘other LSMs’

• Corrected

• Last Paragraph in Section 1: ’17 FLUXNET’----Add one sentence to briefly introduce FLUXNET

• Following the reviewer suggestion, the sentence was changed to:  “As in Best et al., observed data from 17 stations of the FLUXNET global network of micrometeorological flux measurements.

• Section 2.1.2: How Quality Control (QC) is defined and calculated? How to interpret the values of QC?

• The quality control flags mentioned in section 2.1.2 are the quality control flags part of the products. We detailed the exact quality control flags usage in the filtering to allow reproducibility of the results. The explanation of the meaning of each QC value used was included  in section 2.1.2. 

• Section 2.2: ‘the canopy resistance (rc) [23]’----‘the canopy resistance (rc) [23] is calculated as’

• Changed

• The table caption for table 2 is not correct.

• Thank you for calling our attention to this. The correct caption was included “Model simulations acronyms and detailed configuration”

• Table 3 and Table 4: Is there a way to transform the tables into figures? Figures should be more visually clear to the readers instead of stacks of numbers.

• Following reviewer #1 comment, the “bounding box” columns were removed. This helps to improve the readability of the table. The numbers could be displayed as bar plots (as we show the individual scores in supplementary material e.g. Figure S18).  Since there is no expected relation between LAI/Albedo between the different stations, the bar plots are also difficult to visualize.