Relevance of near-surface soil moisture vs. terrestrial water storage for global vegetation functioning
Abstract. Soil water availability is an essential prerequisite for vegetation functioning. Vegetation takes up water from varying soil depths depending on the characteristics of their rooting system and soil moisture availability across depth. The depth of vegetation water uptake is largely unknown across large spatial scales as a consequence of sparse ground measurements. At the same time, emerging satellite-derived observations of vegetation functioning, surface soil moisture and terrestrial water storage, present an opportunity to assess the depth of vegetation water uptake globally. In this study, we characterise vegetation functioning through the Near-Infrared Reflectance of Vegetation (NIRv), and compare its relation to (i) near-surface soil moisture from ESA-CCI and (ii) total water storage from GRACE at the monthly time scale during the growing season. The relationships are quantified through partial correlations to mitigate the influence of confounding factors such as energy-related variables. We find that vegetation functioning is generally more strongly related to near-surface soil moisture, particularly in semi-arid regions and areas with low tree cover. In contrast, in regions with high tree cover and in arid regions, the correlation with terrestrial water storage is comparable to or even higher than with near-surface soil moisture, indicating that trees can and do make use of their deeper rooting systems to access deeper soil moisture, similar to vegetation in arid regions. In line with this, an attribution analysis that examines the relative importance of these soil water storages for vegetation reveals that they are controlled by (i) water availability influenced by the climate and (ii) vegetation type reflecting adaptation of ecosystems to local water resources. Next to variations in space, the vegetation water uptake depth also varies in time. During dry periods, the relative importance of terrestrial water storage increases, highlighting the relevance of deeper water resources during rain-scarce periods. Overall, the synergistic exploitation of state-of-the-art satellite data products to disentangle the relevance of near-surface vs. terrestrial water storage for vegetation functioning can inform the representation of vegetation-water interactions in land surface models to support more accurate climate change projections.
Prajwal Khanal et al.
Status: open (until 08 Jul 2023)
- RC1: 'Comment on egusphere-2023-770', Andrew Feldman, 23 May 2023 reply
Prajwal Khanal et al.
Prajwal Khanal et al.
Viewed (geographical distribution)
Khanal et al evaluate the relative controls of surface and deeper soil moisture on vegetation using mostly satellite data from CCI soil moisture and GRACE terrestrial water storage. They argue that surface soil moisture controls vegetation generally more than deeper moisture stores depending on climate, but deeper moisture stores increase their control in drier months. I have a positive outlook on the study. It evaluates an unexplored gap about the depth of vegetation water use which remains a highly uncertain part of the biosphere. The use of satellite retrievals was highly appropriate for the (nicely posed) objectives. I do think the study needs more work. I have several concerns about the statistical analysis and interpretation of results that should be addressed. I also think more context needs to be added about the uncertainty of the depth of representation of GRACE and CCI and how their correlation confounds interpretation of results. Because of these issues, I think some of the current conclusion arguments extend beyond what we can say from the analysis and may need to be tempered somewhat. Nonetheless, even with the tempered arguments, the study is very insightful and can be a great contribution to the community. See comments below.
L71-79: It would be helpful to expand on what depths TWS represents from previous literature. I’ve seen studies noting 1-3m because this is where most of the water variations are that GRACE can detect. This point is uncertain but it would be good to lay out the previous knowledge for the reader
L83: really great questions
L139: Since GRACE is 0.5 degrees, I recommend conducting the analysis at 0.5 degrees. Some errors may otherwise arise in attribution with higher resolution datasets with the SHAP analysis
L146: are the results sensitive to this threshold? I imagine this could greatly reduce time periods of the year for arid regions
L154-155: It would be helpful to show the regression equation(s) for how this was done. I think this is equivalent to a multiple regression with NIRv as the explained variable and the climate variables as explanatory variables
L157: Both TWS and surface soil moisture have to be negative correlations or insignificant? What if only one of them shows a positive and significant correlation?
L162: More detail is needed for how these dry months are determined. Are these the driest of the growing season months within a pixel?
Figure 1: It would be helpful to show the pdf of the spatial distributions so the values can be more easily viewed
Line 206-211: Can this be shown in a figure?
L216: I think Figure S2 would be valuable to show in the main text. This provides a lot of context especially with regard to my major comments
L291: Can max rooting depth be included in the analysis? The Fan et al 2017 PNAS dataset can be used.
L349-351: I think this argument extends beyond what the correlations say. This conclusion is only based on the small correlation differences. Maybe a more specific significance test is needed here. See my major comments.
L351-353: What if surface soil moisture and TWS are correlated and both are correlated with vegetation behavior? This leaves open the possibility that vegetation is only accessing shallow moisture but shallow moisture is also correlated with deep moisture, which confounds the results. I don’t deny that vegetation is accessing TWS but the conclusion here may need some more evidence based on the current study setup. See my major comment.
L355-359: awesome finding. Well done.
L399: This study is now published in WRR. Please reference that instead of the preprint. https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2022WR033814