the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Apr 2026
Spatiotemporal dynamics of shear stress partitioning around vegetation in a sparsely vegetated dryland
Abstract. Vegetation is the dominant regulator of wind erosion and dust emission in drylands. Drag partition theory is typically used to quantify how vegetation reduces surface shear stress, but lateral flow acceleration along plant sides and the resulting spatiotemporal patterns of shear stress redistribution around individual plants in the field remain understudied. Here we quantify the surface shear stress ratio, its spatial and temporal variability, and lateral flow acceleration along the side of a shrub across a range of wind magnitudes and phenological phases. The spatial variability of the surface shear stress ratio exceeded that reported previously, whereas the temporal variability was 75 % of the spatial variability. Along the shrub side, surface shear stress ratio was independent of wind speed and exhibited a statistically significant decreasing trend with foliar growth while maintaining pronounced spatial heterogeneity, including localized shear amplification near shrub margins. The shape and position of the acceleration zone was broadly consistent with prior studies, although uncertainties remain regarding how vegetation type and dimensional traits influence its extent and form. Incorporating phenology-dependent, yet wind speed-independent, shear stress redistribution and spatiotemporal variability into drag partition schemes provides a pathway toward improved process-based representation of vegetation-wind interactions in wind erosion models. This refinement could improve predictions of the timing, location, magnitude, and frequency of aeolian sediment transport and dust emission, including low-intensity events when above-canopy winds are near or below conventional thresholds.
- Preprint
(2373 KB) - Metadata XML
-
Supplement
(851 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1635', Anonymous Referee #1, 28 Apr 2026
-
RC2: 'Comment on egusphere-2026-1635', Anonymous Referee #2, 08 Jul 2026
Manuscript reference- egusphere-2026-1635
The authors quantified the surface shear stress ratio, its spatial and temporal variability, and lateral flow acceleration along the side of a shrub across a range of wind magnitudes and phenological phases. Incorporating phenology-dependent, yet wind speed independent, shear stress redistribution and spatiotemporal variability into drag partition schemes provides a pathway toward improved process-based representation of vegetation-wind interactions in wind erosion models. The paper is well written. The results are well presented and discussed. There are minor comments as follows:
Comments:
Graphical abstract.
Shrub center has been indicated by in the legend but has been missed in the figure.
Line 197-201. This paragraph seems to be a part of methos section, and it is better to be moved to method section where figure 2 is placed.
Line 203:
In “The Mann-Kendall trend test (p <0.05) indicated a statistically significant decrease trend” the location of (p <0.05) is not fine, rephrase as “The Mann-Kendall trend test indicated a statistically significant decrease trend (p <0.05)”
In “showed a statistically significant decrease trend across phenological phases in both years (Fig. 6).” include P value after (significant decrease)
Line 210:
Include P value for “significant increase trends”
Line 266 and Line 275:
No statistically significant relationships were found between U and 𝑅′ (p <0.05, R2 >0.1).
No statistically significant relationship was found between 𝑅′ and U for p <0.05, R2
>0.1
“p <0.05” should be “p >0.05”
I suggest including P values everywhere statistically significant trend, decrease or increase has been indicated.
Figure 5:
This figure shows spatiotemporal trends of the shear stress ratio across phenological. The phrases are from Feb 14 to July 3 for 2023 and from Feb 19 to June 2 for 2024 which seems slightly inconsistent with Feb 14 to July 7 for 2023 and Feb 12 to June 20 for 2024 in Paragraph 200.
Line 283: “ (Fig. 9a–b)” is probably “(Fig. 10a–b)”
Citation: https://doi.org/10.5194/egusphere-2026-1635-RC2 -
RC3: 'Comment on egusphere-2026-1635', Anonymous Referee #3, 14 Jul 2026
The manuscript “Spatiotemporal dynamics of shear stress partitioning around vegetation in a sparsely vegetated dryland investigates” an interesting and relevant topic with the potential to provide valuable insights into the spatial and temporal variability of surface shear stress around a typical shrub vegetation. The dataset is unique and the field measurements are ambitious; however, several aspects of the manuscript require clarification and strengthening before the conclusions can be fully supported. In particular, the study would benefit from clearer articulation of the research objectives and hypotheses, more detailed descriptions and justification of the methodology and calibration procedures, and additional discussion of key assumptions underlying the shear stress measurements and their interpretation. There are also opportunities to improve the statistical treatment of the data, and reduce potential issues associated with pseudo-replication to overfit regressions. Finally, revisions to the figures, terminology, and overall organization would substantially improve readability and help readers better interpret and generalize the findings. The comments below are intended to help strengthen both the scientific rigor and the clarity of the manuscript.
Page 2 : Line 40 missing verb in sentence – suggest rephrasing : “speeds at”
Page 2 : Line 42 ??well recognized?? by wind tunnel studies then or simulations or both? Maybe worth specifying
Page 3 : Line 71-72 Are any of these are omnidirectional horizontal wind speeds (cup anemometer or rotated 2D or 3D) versus wind vector measurements from LDA or PIV that could be ignoring spanwise flow?
Page 4 : Line 88 This is a ratio with the denominator fixed and so this should be specified as surface shear stress redistribution
Page 4 : Line 90-95 Expecting a series of hypotheses here or maybe objectives/research questions but these are the results. Perhaps reword to better suit a research article model.
Page 4 : Line 103 What pressure transducer was used? Are there inter-sensors calibrations made to estimate their inter-sensor error?
Page 4 : Line 106 Is a frequency of 1 Hz sufficient for resolving temporal variability when compared to lab/model experiments?
Page 4 : Line 108 Horizontal error of 19 cm: do you mean +/- 19 cm? These seems a lot for a ~1 m tall shrub
Page 4 : Line 110 What is the variation in density of these? Scattered grasses and forbs along the fetch? over the measurement period? Is this something that could influence the total shear stress
Page 5 : Line 116 Pictures show four sonic anemometers installed quite close to each other that could cause interference. Assuming that the most upwind is the one used in this study?
Page 5 : Line 117 What is the path length and therefore maximize eddy size that these sensors can capture installed only 1 m above the surface?
Page 5 : Line Fig. 1 If a linear interpolation was targeted why are the sensors not installed on a standard grid?
Page 5 : Line Fig. 1 What about the different years - are all these present for both years? Helpful to have some different colors or symbols to identify the different years
Page 5 : Line Fig. 1 Why wasn't the full shadow area targeted as much as the upwind area?
Page 7 : Line Fig. 4 Indicate which points are from what method
Page 7 : Line 138 Under what conditions were they calibrated in the SENSE tunnel - what kind of surface fetch etc?
Page 7 : Line 138-140 Aren't the PI-SWERL shear velocity estimations based on wind tunnel calibrations to the RPM? Are there overlaps between the wind tunnel and PI-SWERL calibrations for continuity? More explanation/discussion is needed here to justify this approach and it is crucial to the estimation of main variable used in this study
Page 7 : Line 144 The R^2 and RMSE are over and underestimated respectively due to the pseudo-replications of 10+ measurements made at the same U*s. These values should be averaged before the power law is applied to reduce over fitting and inflated confidence in the calibration.
Page 8 : Line 150-151 This wind speed corresponds to what U* value? How does sediment transport impact the results knowing that this windspeed filter likely results in below and above transport threshold conditions to be included?
Page 8 : Line 157-158 A better explanation on how the authors can justify that a 1 m high shear stress measurement is representative of the total shear stress of the surface boundary layer in 1 m high sparsely vegetated surface. Although the surface values are all normalized by this same value the likely underestimation from placing the instrument in the inertial sublayer along with larger variability will provide erroneous estimates of the SSR.
Page 8 : Line 167 Again this is not likely representing the total shear stress - justification needed.
Page 8 : Line 172 ?5-min samples? are these not averages?
Page 8 : Line 176 This is indeed a location minimally influenced by the shrub of interest, but is it representative of the average surface U*? If when takes the ratio of 1 as the threshold for sheltering/acceleration it assumes this is the average or median value (depending on the distribution) of the surface values, not just a value without influence by the shrub. More explanation is needed on why this value was chosen.
Page 9 : Line 184 ??we filtered the data to require that 8,10 and 18 sensors??: But to minimize spatial bias it is not only the number of sensors, but where they are placed - which row and what distance from the shrub. Where can this information be found?
Page 9 : Line 185 Filtered to a narrower wind direction narrower than what values? If you do not need to compare then you can omit the comparative word.
Page 9 : Line 192 Is it not more informative to compare these values to the freestream windspeed like in other studies?
Page 9 : Line 199-200 This is quite a wide variation and relatively small number of 5 minute average samples for the 10 and 18 sensors for 2023 and 2024 respectively.
Page : Line 706 87 and 477 correspond to roughly 70, 8, and 47, 5 minute periods in 2023 and 804, 182, and 443 corresponds to only 44, 10, and 24 5-minute periods in 2024 with the information provided.
Page 9 : Line 203 Mean values of R' or R in space or in time? They appear to be the 5 minute average samples of U*t divided by U* for each sensor and for each 5 minutes (not explained in methods 2.4). If the goal is to identify temporal trends, would it make sense to take instead the daily average of all sensors to again minimize what seems like a pseudo-replication to boost trend significance
Page 10 : Line Fig. 5 Figures could be improved by placing the legend in a single box on the first figure providing letters for the subplots, improving the caption to point out which figures are for which years
Page 11 : Line Fig. 6 If the goal is to look at differences as a result of greening, daily averages or indeed an anova on the values regrouped by the three greening phases would be more appropriate. Here the significance of the trend is masked/emphasized by the clumps of values that just identify a change over the greening phases versus a rate of change.
Page 11 : Line Fig. 6 Subplot letters are needed here
Page 11 : Line 217 Remove or at least replace the X and Y variables with the actual variables
Page 11 : Line 225 Why is this not also calculated for the temporal variability?
Page 12 : Line 234-240 Suggest moving to Annex or supp material and explaining in methods that there was similar results between years when not accounting for new sensors. Although it is curious that if indeed there are no differences that the two years of data are not amalgamated in the regression analyses. Also, only S2 is referenced in manuscript. Suggest perhaps combining figures.
Page 12 : Line 241 Are these regrouped by downwind positions or treated as individual? Where are these results shown (table/figure)?
Page 13 : Line Fig. 8 Sensor location names need to be added here to help guide the reader while following the text
Page 13 : Line Fig. 8 Are the phases a1/a2 b1/b2 and c1/c2? They are not referred to this elsewhere if that is the case and overlap with the sensor position naming convention adding confusion
Page 13 : Line 251 Based on the discussion below perhaps expressing the values in Figure 8 as the percent difference from the upwind value is more helpful for the discussion? Or a combined graphic with colors as absolute R and contours as percent difference
Page 13 : Line 252-253 First mention that this is where (some) of the additional eight sensors are located ? this info needs to be in the methods.
Page 14 : Line 269 Used relationship and regression before and now trend to explain these results - perhaps try to utilize the same terminology.
Page 15 : Line Fig. 10 Suggest providing a qualitative colorbar instead of a continuous one that is divergent from the value of zero if the intention is to show the negative or positive changes
Page 15 : Line 283 These locations should be referenced relative to the axes provided (normalized distance) to allow for transferability versus project specific nomenclature.
Page 15 : Line 288 So the figure includes outliers or not? Not clear from this phrasing and without letter labeling it is impossible to tell. Suggest removing outliers on result figures, while providing raw results in supp mat.
Page 15 : Line 289-290 Not sure there is enough evidence here alone to suggest that the wind forcing drives growth.
Page 15 : Line 298-299 Again here the relative description of changes to the row letters is not helpful for the reader nor for transferability of the results
Page 15 : Line 304 Is this just a function of windspeed then? Perhaps the data should be partitioned by below and above threshold vs an arbitrary windspeed to better reflect logistical challenges (scouring).
Page 16 : Line Fig. 11 Not clear what the subfigure alphanumeric references here as well. Suggest providing (a) through (f) captions
Page 16 : Line Fig. 11 The annotated row names on the x-axis are somewhat helpful but from figure 1 do not necessarily seem to be properly placed as they do not follow an actual grid spacing. Therefore my suggestion is to use normalized distance rather than row names anywhere in the manuscript
Page 18 : Line 354-355 Could this lack of relationship also be due to the U reference being in the sublayer and not above canopy flow making R' and U correlated
Page 20 : Line 402-404 I am not sure how this was elucidated by the present results. Scouring effects were not described beyond a logistical problem for sensor measurement.
Page 20 : Line 411 Likely due to the reduced spatial coverage of the measurements in this study than it being due to the study being conducted in a natural environment.
Citation: https://doi.org/10.5194/egusphere-2026-1635-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 228 | 76 | 22 | 326 | 70 | 26 | 30 |
- HTML: 228
- PDF: 76
- XML: 22
- Total: 326
- Supplement: 70
- BibTeX: 26
- EndNote: 30
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript addresses the spatiotemporal dynamics of surface shear stress partitioning around individual shrubs in drylands. They use high-quality field measurements with dense Irwin sensors across two years and three phenological phases, providing direct observations of shear velocity that are rare in the literature. The data are well calibrated, the analyses are statistically sound, and the main findings are robust. The primary finding is that the surface shear stress ratio R’ is independent of wind speed but shows a significant decreasing trend with foliar growth—challenges conventional drag-partitioning theory and has important implications for process-based wind erosion modeling. While the study is limited to a single shrub, it provides a valuable empirical dataset and a clear conceptual advance.
To my knowledge, this is the first study to deploy a dense array of Irwin sensors to directly measure u∗s (not just wind speed) around a natural shrub over an extended period (~11 months).
The demonstration that is phenology-dependent but wind-speed-independent is a significant conceptual contribution. It means that in sparsely vegetated landscapes, the spatial pattern of erosion potential is governed more by vegetation structure than by instantaneous above-canopy wind forcing. This should prompt a re-evaluation of assumptions in many wind erosion models. I like this result.
The clear finding that spatial variability in exceeds temporal variability (by ~30-40%) is highly relevant for model parameterization. It underscores the necessity of accounting for fine-scale spatial heterogeneity, not just temporal averages, when predicting sediment transport.
All conclusions are drawn from measurements around a single Prosopis glandulosashrub (~1 m tall). The shape, extent, and magnitude of the acceleration zone, as well as the absolute values of and its spatial variability, are likely influenced by species-specific traits (morphology, porosity, flexibility) and scaling (height-to-width ratio). The crown diameter was nearly twice the height in this study, which may explain differences with prior studies using cylindrical elements.
The figures could be refined; they appear rather rough and unpolished overall.
The analysis is restricted to wind speeds between 4 and 10 m/s at 1 m height. The conclusion that is independent of wind speed is robust within this range. Is it sound?
The experiment is based on only one individual shrub of a single species and only the dominant wind direction (225°±30°). The authors must clearly and explicitly state that their conclusions are site-specific and shrub-specific, and cannot be broadly generalized to other vegetation types, morphologies, densities, or wind directions. Overstatements about “shrubs in drylands”. The study only focuses on one single shrub, not “vegetation” in general, so the title might be something like “Spatiotemporal dynamics of shear stress partitioning around a single shrub in a sparsely vegetated dryland” or “Shear stress partitioning around a single shrub: spatiotemporal responses to wind magnitude and phenology in a dryland”…
I did not see the quantitative measurements of shrub morphology (canopy width, height, porosity, branch density).
Figures 5 and 6 visually appear to show rather weak trends, and even if so-called “objective results” are derived from non-parametric statistical tests like the Mann-Kendall test, they may not truly explain anything substantial.
The conclusions section largely repeats findings already in the abstract.
Th “first” are slightly overstated, appearing many times in the text.
The 4 m s⁻¹ wind threshold and 5-minute averaging interval are acceptable, but the authors should briefly justify why this choice does not bias the estimation of shear velocity.