the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Apr 2024
Critical soil moisture detection and water-energy limit shift attribution using satellite-based water and carbon fluxes over China
Abstract. Critical soil moisture (CSM), a tipping point of soil moisture (SM) when evapotranspiration (ET) begins to suffer from water limitation, is essential for the vegetation state and corresponding land‐atmosphere coupling. The water and energy-limited regime in various biomes and climates shifts under global climate change. However, detecting CSM and attributing water-energy limit shifts to climate and ecosystem variables are challenging as in-situ observations of water, carbon fluxes, and SM are sparse. In this study, CSM derived from two satellite-based methods were assessed over China: the difference between the correlation between SM and ET and the correlation between vapor pressure deficit (VPD) and ET (correlation-difference method) using four satellite-based ET products; the covariance between VPD and gross primary production (GPP) (VPD-GPP-SM method) using four satellite-based GPP products. The ET and GPP products were Penman-Monteith-Leuning (PML) ET and GPP, Global LAnd Surface Satellite (GLASS) ET and GPP, Collocation-Analyzed Multi-source Ensembled Land Evapotranspiration (CAMELE) ET, Surface Energy Balance Algorithm for Land evapotranspiration in China (SEBAL) ET, Two-Leaf light use efficiency model based (TL) GPP, and SIF-based GPP (GOSIF). At flux sites, satellite-based ET and GPP were evaluated by the eddy covariance technique, and CSMs derived from the site-level GPP and ET data were evaluated by the evaporative fraction-soil moisture (EF-SM) methods. Our study revealed that the performance of ET, GPP, and CSM at the site scale demonstrated reliable results and applicability to regional scales. The intercomparison of CSM from multi-source ET and GPP datasets across China indicated their consistency and robustness. Generally, CSM decreased from southern to northern regions of China and decreased with increasing layer depth, particularly in the Tarim Basin and Haihe River Basin. Areas characterized by clay-rich soils (e.g., 0.39 m3/m3 using GOSIF GPP and 10 cm depth SM), and forests (e.g., 0.37 m3/m3 using GOSIF GPP and 10 cm depth SM), and located within the Pearl River Basin and Southeastern River Basin displayed a relatively high CSM. Decreased energy limitations in western and southern regions in June–September over the period 2001–2018 were associated with increasing ET and decreasing precipitation, respectively. The findings highlight the variability in CSM and its primary determinants, offering valuable insights into the potential water limitation on both ET and GPP processes under comparable SM circumstances.
- Preprint
(4518 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on hess-2024-105', Hsin Hsu, 18 Jun 2024
This study applies multiple methods to identify critical soil moisture (CSM) that separates water- and energy-limited regimes using several satellite-based data and in-situ observations with a specific spatial scope. Then, it explores the factors that dominate the variations using a feature regularization technique. I support this study as I think their analyses on CSM and the determinant factor advance the science on the water and energy cycle over land. However, I think the readability of the paper and the description of the analyses should be further improve. Moreover, I have concerns about the confidence in the CSM estimates. I suggest a major revision. Please see my comments below:
I do not wish to remain anonymous – Hsin Hsu
Major Comments
- The article is very hard to read because the amount of abbreviation is overwhelming. I had to look up what an abbreviation stands for multiple times in just one sentence as they are from different categories (variables, locations, names of in-situ sites, land cover types, algorithms, statistical parameters, data products, etc.). I suggest retaining the full names for land cover types and some algorithmsin the writing, as many do not occur frequently in the paper. The authors could also consider separating the abbreviations by different systems. For example, use Greek alphabets for statistical parameters and italics for variables.
The main methodologies used in this study may need abbreviations due to their lengthy names (e.g., Corr(ET,VPD) - Corr(ET,SM)). Sometimes they occur many times in one paragraph (or even one sentence), but the key information separating different correlation-difference methods is the second variable used in the first correlation calculation. The authors could consider modifying the notation of Denissen et al. 2020 to define:
ΔCorrVar = Corr(ET,Var) - Corr(ET,SM).- Most of the cited work on critical soil moisture and regime examination is published before 2022. There are many new aspects of regimes and CSM since 2023. Not required to reference butthe authors could consider integrating these recent studies:
New method for calculating CSM based on satellite data:
Fu et al. (2024). Global critical soil moisture thresholds of plant water stress. https://doi.org/10.1038/s41467-024-49244-7Global estimation of CSM based on soil moisture dry-down framework and an index to quantify vulnerability:
Dong et al. (2023). Land Surfaces at the Tipping-Point for Water and Energy Balance Coupling. https://doi.org/10.1029/2022WR032472In the abstract (line 11), author mentions that regimes can shift under climate change. This is not discussed in the introduction:
Hsu, H., Dirmeyer, P.A. (2023). Soil moisture-evaporation coupling shifts into new gears under increasing CO2. https://doi.org/10.1038/s41467-023-36794-5
Hsu, H., Dirmeyer, P.A. (2023). Uncertainty in Projected Critical Soil Moisture Values in CMIP6 Affects the Interpretation of a More Moisture-Limited World. https://doi.org/10.1029/2023EF003511
Duan et al. (2023). Coherent Mechanistic Patterns of Tropical Land Hydroclimate Changes. https://doi.org/10.1029/2022GL102285.- The robustness of each technique for estimating CSM is not described:
From Figure 4, it seems that CSM can be identified in almost all grid cells by the covariance method. Is this also true for all other methods? If not, what is the rate of agreement on the existence of CSM among all the methods?
Do all methods for estimating CSM have significant tests? For example, when using the SM-EF method, it is common to use three different linear models (flat line, a positive slope line, and a linear-plus-plateau) and select the best one based on the Bayesian information criterion (BIC). If flat-line regression or positive-slope regression outperforms linear-plus-plateau regression, CSM should be considered as not identified. This procedure is also used for detecting CSM by the soil moisture-drydown framework.
When using the correlation-difference method, if there is more than one SM value where the correlation-difference is zero, which SM value is identified as CSM? In Figure 3b, the red line locates at the wetter SM value when the correlation-difference is zero, but blue-greenline locates at the drier SM value when the correlation-difference is zero. This should be clarified. Additionally, I assume that if either correlation is not statistically significant when calculating the correlation before taking the difference, CSM should also be treated as not identified. Is this the case in this study?
- When examining the alignment of CSM between different methods, statistical significance is needed. I recommend a Chi-square test as it can address scenarios involving categorical data: comparing rates or proportions between two groups when the outcome is a binary variable, such as negative and positive outcomes. In this specific case, categorical data represent soil moisture (SM) values tagged as "drier-than-CSM" and "wetter-than-CSM". So, a Chi-square test can be used to compare the proportions of SM values below and above CSM between two sets of variables or groups (obtained by different methods). If there are significant differences, it means the CSM is different.
- Figure 5 seems problematic. The CSM is extremely well aligned among different ET products. However, I assume the spread in temporal variation among ET productsover many locations based on Figure 2a, where the correlation of each product’s ET to in-situ data can be very different. Does that not lead to a different CSM estimate? The authors could provide some supporting information to justify the consensus, which looks too good. The tick labels on the y-axis in each bar chart are incorrect. The bars are spatial means, so error bars should also be provided.
- In Figure 6, the CSM among different SM layers is also extremely well aligned (if I interpret it correctly). This makes me doubt the reliability of SM in deeper layers. I assume SM-EF could be decoupled in deeper soils if roots do not reach that deep in some places, so there should be some inconsistency in CSM values. In either case, I suggest the authors provide additional analysis to examine soil moisture at deeper layers(maybe taking some grid cells for example and put as supporting information) and discuss the uncertainty of using these data products in the discussion. For example, some ET products are estimated; are the sources of input to derive ET independent of each other for all of them? Does method to derive SM at different layer inherently lead to consistency in CSM?
- How does the author determine the set of input features for the ridge regression and why air temperature and VPD are not considered in the analysis?
Minor Comments
- Line 213: should be "ea" not "ea."
- Line 122: The term REddyProc is not explained nor mentioned elsewhere.
- I suggest putting the unit of variables in every chart.
- Is ΔCorr calculated monthly between June and September and then an annual mean is obtained?
- The word “Slope” in figures 7 and 8 is confusing. Is this a temporal trend? What is the statistical significance?
- Does the author perform cross-validation or bootstrapping for ridge regression?
- AC1: 'Reply on RC1', yi liu, 28 Oct 2024
-
RC2: 'Comment on hess-2024-105', Jingwei Zhou, 03 Oct 2024
General Comments:
This manuscript presents an analysis of critical soil moisture (CSM) across China using multiple satellite-based datasets of evapotranspiration (ET), gross primary production (GPP), and soil moisture (SM). The authors apply two methods to detect CSM - a correlation-difference approach and a VPD-GPP-SM covariance approach. They evaluate the spatial patterns of CSM across different land cover types, soil textures, and regions of China, and analyze the factors driving shifts between water and energy-limited regimes.
Overall, this study represents a substantial contribution within the scope of HESS. The use of multiple satellite datasets to examine CSM at large scales is novel and provides new insights into water and energy limitations across diverse landscapes in China. The methods are generally sound and though the results are discussed not comprehensively in the context of related work. The manuscript is in general structured, though some sections could be more concise and many textual improvements might be needed. I will give a major revision for this work.
Specific Comments:
- The methods section is quite detailed, which is good for reproducibility. However, some of the dataset descriptions are too redundant right now and could potentially be shortened or moved to data availability statement to improve readability of the main text. For example, subsections such as 2.1.2, 2.1.4, 2.1.5.
- The manuscript uses too many abbreviations from various categories (region name, physical variable, mathematical methods, etc), making it difficult to read. Consider retaining full names for less frequently used terms to improve clarity.
- The evaluation of satellite-based ET and GPP products against flux tower data (Section 3.1) is valuable. However, more discussion (section 4) of the implications of biases in these products for the CSM analysis would strengthen the paper.
- The comparison of CSM detection methods at flux tower sites (Section 3.2) is an important component. The authors could consider adding a quantitative metric of agreement between methods and also a statistical test to assess the significance of the agreement to supplement the qualitative comparisons.
- The high level of alignment in CSM estimates among different ET products and soil moisture layers is surprising given the potential variability in these datasets. Additional analysis or explanation is needed to justify this consistency and discuss potential uncertainties. Moreover, the consistency in CSM values across different soil depths raises questions about the reliability of deeper soil moisture data. Consider providing additional analysis of deep soil moisture to showcase the possible difference between the soil layers and discussing the uncertainties associated with these measurements.
- The attribution analysis of water/energy limit shifts (Section 3.4) provides valuable insights. The authors could consider expanding on the implications of these findings for water resource management or ecosystem responses to climate change, perhaps in discussions part.
- The discussion section effectively contextualizes the results within existing literature. However, it could be strengthened by more explicitly addressing the limitations of the approach used in this manuscript and potential future research directions.
Some textual suggestions:
Line 10: “suffer from water limitation”, this kind of metaphor is not suitable I assume, please change it to some words that are not for humans.
Line 14: Put “were assessed over China” before “derived” in the sentence to let the colon directly connecting the following methods.
Line 24: Maybe change the sentence to “Through intercomparison, CSM from multi-source ET and GPP datasets across China is found to be consistent and robust.”
Line 34: Please don’t repeat sentences in your manuscript, this one is the same with that in your abstract, please revise either of them.
Line 44: “customary” and “matric”, change them to some others relatively commonly used.
Line 98: “comparability” is not common in papers. Some noun forms of words are not common to be used even in academic world. Please consider adjective forms and revise the relevant sentence or use other nouns.
These are just what I roughly found, please read the text thoroughly after revision and also consider a text revision service.
Figure 1: Consider changing the colors of forest and grassland to make them easier to differentiate as the landcovers you have are not so many
Figure 5: Consider making it into two columns
Figures 3 and 4: consider use noun as the subject in the titles rather than using verbs
Line 199: “9-day moving windows”?
Citation: https://doi.org/10.5194/hess-2024-105-RC2 - AC2: 'Reply on RC2', yi liu, 28 Oct 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 401 | 107 | 76 | 584 | 18 | 21 |
- HTML: 401
- PDF: 107
- XML: 76
- Total: 584
- BibTeX: 18
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
