Recent Changes in the Low-Level Jet along the Subtropical West Coast of South America

Round 1

Reviewer 1 Report

General comments:

1. line 174 - threshold 10 m/s - which direction? only meridional wind component? if we have E=4 m/s and V=9.5 m/s in is not a CLLJ? If the coast line have 340- 30-40° northward direction? I think you must calculate all statistics for sector for example 345-30° with absolute wind velocity more then 10 ms. Now you have only a few events per year, may be this can be grow up.
2. Section 2 - I think the structure needs to be changed. You need to make a section "Wind quality assessments", which include comparison method and results of comparison. Please make it more clear which data you use for find CLLJ events. Daily average Meridional wind speed?
3. "Six-hour SLP data from ERA5 were interpolated onto a 2.5° × 2.5° horizontal grid" - this is a method which provided 20 years ago. You have a high resolution 0.25 reanalysis and interpolate it to a 2.5°?? It is a wrong I think. May be you can get modern method of cyclone tracking?
4. If I understand correctly,  you use an average daily wind speed of more than 10 m / s to identify CLLJ. In Figure 4, the top panel, we see the presence events > 10 m/s in measurements and the absence of this data in the reanalysis. Your research can be recognized as relevant? You must explain this

Minor comments:

1. figure 1 title - v10 and SLP did not presented earlier
2. ERA5 - why are you use ERA5? it is any publication which confirm that ERA5 better Then NCEP/CFSR or other reanalysis for your study area? please insert the links about wind quality era5 for your region
3. figure 2 - please show the zonal wind component U, because it is significant to understand direction. On the pressure maps I see that it is different direction during one case of CLLJ
4. Table2 - please check the table, looks very strange when the Bias greater then MSE in several cases. 
5. "semi-permanent" may be better to use "Quasi-stationary"
6. figure 4 - axis limits for v10 can be reduced, for Faro Corona -35 m/s is not needed
7. "These metrics indicate that reanalysis data and in situ measurements are in close agree-235 ment for V10" - this is wrong, because your investigation based on absolute threshold 10 m/s, and R correlation is not a criteria for this. MSE is more suitable  and 4-6 m/s is not good...
8. Big MSE in your comparison can be caused by unrepresentative stations. You can use the wind from scatterometers over the ocean to show that the renanalysis reproduces the wind well (although there is a problem that satellite data is assimilated in reanalysis, but it's better than when the weather station is near the mountains)
   
9. figure 6e - may be better to use Probability, when v10 higher 10 m/s
10. all figures - if you use Lattitude (S°) - you must to use absolute values of lattitudes, without "-"
11. figure 7 - please add the R2 on linear trends.

Author Response

Response to Reviewer 1 Comments

We thank the reviewer for his/her time and constructive comments. We carefully read each of the comments and suggestions and addressed them.

 General comments 

1. Line 174 - threshold 10 m/s - which direction? only meridional wind component? if we have E=4 m/s and V=9.5 m/s in is not a CLLJ? If the coastline has 340- 30-40° northward direction? I think you must calculate all statistics for sector for example 345-30° with absolute wind velocity more then 10 m/s. Now you have only a few events per year, may be this can be grow up.

R: We agree with the reviewer, CLLJ winds are basically coastal parallel as they are forced by alongshore pressure gradients. Although the Chilean coast is approximately straight north-south we followed your advice and   calculated the alongshore wind speed (V_a) in the revised manuscript which resulted in slight changes in figures 4 to 7, figure 9 and table 2. Nevertheless, the main findings showed in the original manuscript remain unchanged in this revision.

The 10 m/s value, as any threshold is somewhat arbitrary, but it is chosen from wind speed distribution reported on previous studies (Garreaud and Muñoz, 2005, Rahn and Garreaud, 2014). However, in Figure 7 of the original manuscript, there was a mistake showing the Number of days and Number of events per month. This was corrected in the revised manuscript where numbering in figure 8 now effectively correspond to events per year. 

• Garreaud, R. D. & Muñoz, R. C. The low-level jet off the West Coast of Subtropical South America: structure and variability. Mon. Weather Rev. 133, 2246–2261 (2005).
• Rahn, D. A. & Garreaud, R. D. A synoptic climatology of the near-surface wind along the west coast of South America. Int. J. Climatol. 34, 780–792 (2014).

2. Section 2 - I think the structure needs to be changed. You need to make a section "Wind quality assessments", which include comparison method and results of comparison. Please make it more clear which data you use for find CLLJ events. Daily average Meridional wind speed?

R: We are not able to change this structure as we must follow a template with predefined sections. In Materials and Methods section, we have to indicate the datasets used, this is mandatory. Then, we moved the error metrics from subsection 2.2 “Data processing” to the subsection 3.1 now entitled "Wind quality assessments." We think these changes will satisfy your suggestion.

3. "Six-hour SLP data from ERA5 were interpolated onto a 2.5° × 2.5° horizontal grid" - this is a method which provided 20 years ago. You have a high resolution 0.25 reanalysis and interpolate it to a 2.5°?? It is a wrong I think. May be you can get modern method of cyclone tracking?

R: We don´t agree with the reviewer in this point. First, we argued that migratory anticyclones, that force high-wind events along the coast of central Chile, are indeed represented using a spatial resolution of 2.5° since the average diameter of anticyclones in the Southern hemisphere is in the order of 20° (Pepler et al., 2019). Second, the method of tracking provided 20 years ago is still used in current publications in high impact journals (see some examples below) showing that it has passed the “test of time”. On this basis we feel confident in using this method and resolution. Yet, we understand this may be an issue to the reader (as it was for you) so we have commented on this on lines 185-188 and provide the following references:  

• Sato, K., Inoue, J., Simmonds, I. et al. Antarctic Peninsula warm winters influenced by Tasman Sea temperatures. Nat Commun 12, 1497 (2021). https://doi.org/10.1038/s41467-021-21773-5

• Pepler, A.S., Dowdy, A.J., van Rensch, P. et al. The contributions of fronts, lows and thunderstorms to southern Australian rainfall. Clim Dyn 55, 1489–1505 (2020). https://doi.org/10.1007/s00382-020-05338-8.

• Aguirre, C., Rojas, M., Garreaud, R.D. et al. Role of synoptic activity on projected changes in upwelling-favourable winds at the ocean’s eastern boundaries. npj Clim Atmos Sci 2, 44 (2019). https://doi.org/10.1038/s41612-019-0101-9

• Pepler, A., Dowdy, A. & Hope, P. A global climatology of surface anticyclones, their variability, associated drivers and long-term trends. Clim Dyn 52, 5397–5412 (2019). https://doi.org/10.1007/s00382-018-4451-5

4. If I understand correctly, you use an average daily wind speed of more than 10 m/s to identify CLLJ. In Figure 4, the top panel, we see the presence events > 10 m/s in measurements and the absence of this data in the reanalysis. Your research can be recognized as relevant? You must explain this.

R: Effectively, figure 4 does not contain alongshore windspeeds higher than 10 m/s in the reanalysis. However, this occurs at coastal locations (meteorological stations) where strong winds appear only in the afternoons (i.e. lower daily- mean windspeeds). However our analysis considers windspeeds about 150 km offshore (i.e.CLLJ axis). Indeed, figure 5e in the original manuscript, shows that winds higher than 10 m/s have probabilities of occurrence between 10-30% with temporal and spatial dependence. In the revised manuscript we use winds from QuikScat to compare with the reanalysis ERA5 at the locations of our analysis (following the minor comment #8), where the presence of values higher than 10 m/s is evident (figure 5 in the revised manuscript). 

Minor comments:

1. figure 1 title - v10 and SLP did not presented earlier.

R: The figure caption has been improved.

2. ERA5 - why are you use ERA5? it is any publication which confirm that ERA5 better Then NCEP/CFSR or other reanalysis for your study area? please insert the links about wind quality era5 for your region

R: We added a reference that compares several reanalysis datasets and shows that the ERA5 surface winds exhibit the best agreement, correlating and reproducing the observed variability on a daily timescale.

• Ramon, J, Lledó, L, Torralba, V, Soret, A, Doblas‐Reyes, FJ. What global reanalysis best represents near‐surface winds?. Q J R Meteorol Soc. 2019; 145: 3236– 3251. https://doi.org/10.1002/qj.3616

3. figure 2 - please show the zonal wind component U, because it is significant to understand direction. On the pressure maps I see that it is different direction during one case of CLLJ

R: We added a new panel in figure 2 showing the zonal wind component.

4. Table2 - please check the table, looks very strange when the Bias greater then MSE in several cases.

 R: Thanks for pointing this out. We found an error because we were presenting the Root mean square error. This has been corrected in the revised manuscript.

5. ‘semi-permanent’ may be better to use ‘Quasi-stationary’

R: Good point. The term ‘semi-permanent’ was changed by ‘quasi-stationary’ in the revised manuscript.

6. figure 4 - axis limits for v10 can be reduced, for Faro Corona -35 m/s is not needed

R: Axis limits were changed in figure 4 of the revised manuscript.

7. ”These metrics indicate that reanalysis data and in situ measurements are in close agreement for V10” - this is wrong, because your investigation based on absolute threshold10 m/s, and R correlation is not a criteria for this. MSE is more suitable and 4-6 m/s is not good...

R: We agree with the reviewer.  Following this suggestion and minor comment #8 we added a comparison between ERA5 and QuikSCat data in section 3.1 of the revised manuscript. This allows us to compare ERA5 with observed data in the same locations where we have analyzed winds. However, we maintained the comparison with winds from coastal meteorological stations.   

8. Big MSE in your comparison can be caused by unrepresentative stations. You can use the wind from scatterometers over the ocean to show that the renanalysis reproduces the wind well (although there is a problem that satellite data is assimilated in reanalysis, but it’s better than when the weather station is near the mountains).

R: Following the reviewer´s suggestion we have included a comparison between the wind obtained from the scatterometer QuiKScat and winds from ERA5 in section 3.1 in the revised manuscript

9. figure 6e - may be better to use Probability, when v10 higher 10 m/s

R: We think that reviewer #1 refers to Figure 5e. In this case, the word ‘percentile’ of the x-axis was replaced by ‘probability’ in figure 6e of the revised manuscript.

10. all figures - if you use Latitude (S°) - you must to use absolute values of latitudes, without “-“

R: Thank you for pointing out this redundancy. Only the minus sign for latitudes was kept in all figures.

11. figure 7 - please add the R2 on linear trends.

R: Figure 7 shows the trends in the number of windy days and number of CLLJ events. For this reason, there is not a correlation coefficient associated. However, with the aim of highlighting the significance of these trends, the position of the trend values in the Monte Carlo analysis was included.

Reviewer 2 Report

Section 2

Please check consistency among different reanalysis products to confirm the long-term trend is not specific to ERA5.

Section 3

It is desirable to add some figures showing the differences in anomalies in SLP and V10 between the two periods (2010-2019 and 1979-1988). It makes authors’ claim more convincing.

Section 4

If possible, please add horizontal distributions of anomalies of sea surface temperature and depth of ocean surface mixed layer since you mention the biological impact of the CLLJ.

Lines 421-422 Please elaborate. If you have some supporting evidence, please show it.

Lines 448-450 Please delete these lines.

Author Response

Response to Reviewer 2 Comments

We thank the reviewer for his/her time and constructive comments. We carefully read each of the comments and suggestions and addressed them.

Section 2

Please check consistency among different reanalysis products to confirm the long-term trend is not specific to ERA5.

R: Point well taken. It required substantial extra work, but we found worth comparing ERA5-derived trends with another reanalysis. Among several options, only CFSR has a horizontal resolution similar to ERA5 but it covers a shorter period. Thus, we checked consistency of the long-term trends obtained with ERA5 with those from the CFSR reanalysis for the common period between 1979 and 2010. Results are shown in supplementary material as Figure S1 and exhibit a good agreement in the spatial patterns of trends calculated in both reanalyses, particularly during summer. We added some text on this comparison on lines 382-386. In addition, we added a reference that shows long-term surface wind trends of several reanalysis, highlighting the robustness of changes observed in the southern portion of the Humboldt system.

• Ramon, J, Lledó, L, Torralba, V, Soret, A, Doblas‐Reyes, FJ. What global reanalysis best represents near‐surface winds?. Q J R Meteorol. Soc. 2019; 145: 3236– 3251. https://doi.org/10.1002/qj.3616

Section 3

It is desirable to add some figures showing the differences in anomalies in SLP and V10 between the two periods (2010-2019 and 1979-1988). It makes authors’ claim more convincing.

R: We added two figures in supplementary materials to show anomalies in SLP and V10 during CLLJ events off Punta Lengua de Vaca (30°S) and off Punta Lavapie. Differences in anomalies between the two periods considered in section 3 (2010-2019 and 1979-1988) confirm higher SLP positive anomalies and more intense winds during the last decade. 

Section 4

If possible, please add horizontal distributions of anomalies of sea surface temperature and depth of ocean surface mixed layer since you mention the biological impact of the CLLJ.

R: We agree that it is important to mention possible impacts of the CLLJ changes on coastal ocean (physical and biological variables). Therefore, we included a discussion section, with the relevant references to support our hypotheses. Analyzing the impacts of the CLLJ changes upon ocean variables (including biology), however, is beyond the scope of this paper.

Lines 421-422 Please elaborate. If you have some supporting evidence, please show it.

R: The phrase presented in lines 421-422 of the original manuscript is based on the results obtained by Pepler et al (2019), who compared several reanalysis datasets. For clarifying this, we slightly modified the text highlighting this reference and introduced figure S4 as supplementary material. This figure shows changes in migratory anticyclone density in ERA5 reanalysis when comparing the two periods (2010-2019 and 1979-1988). 

Lines 448-450 Please delete these lines

R: Sorry for the mistake. Lines 448-450 were deleted.

Round 2

Reviewer 1 Report

Good job!

The minor comments:

1. In Keywords I didn't see the "coastal low-level jet", please add
2. Please, Specify  the link to QuikSCAT "http://cersat.ifremer.fr/" I didn't find the detailed description of this data.
3. Please add to Table 2 the total number of data for comparison, for each station or QuikSCAT.
4. Please add the color bar and its title to Fig5.

Author Response

We thank the Reviewer for his/her comments.

The minor comments:

1. In Keywords I didn't see the "coastal low-level jet", please add.

R: Thanks for this comment. Coastal low-level jet has been added as a Keyword. 

2. Please, Specify  the link to QuikSCAT "http://cersat.ifremer.fr/" I didn't find the detailed description of this data.

R: We specify the link to QuikScat data in lines 157-160 of the revised manuscript. 

3. Please add to Table 2 the total number of data for comparison, for each station or QuikSCAT.

R: The total number of data has been added in Table 2

4. Please add the color bar and its title to Fig5.

R: The colorbar and its title have been added in Figure 5.