TL;DR
This study presents the first global map of critical soil moisture thresholds (θ<sub>crit</sub>), the point at which plants experience water stress and evapotranspiration becomes limited. By combining satellite observations of land surface temperature and soil moisture during dry-downs, the researchers found that arid regions have lower thresholds than humid ones, reflecting plant adaptations. This work is crucial for improving climate and water resource models, as it reveals that the frequency of water-stressed days has increased globally over the past four decades.
What Problem Does This Paper Address?
Understanding how much water is available to plants is fundamental to predicting ecosystem health, agricultural yields, and the impacts of climate change. Plants experience water stress when soil moisture drops below a certain point, limiting their ability to take up water and perform essential functions like evapotranspiration. This critical soil moisture threshold (θ<sub>crit</sub>) is a key parameter for modeling these processes. However, a global, observation-based understanding of θ<sub>crit</sub> has been lacking. Previous efforts have relied on simplified models or limited in-situ data, leading to uncertainties in climate projections and an incomplete picture of how terrestrial ecosystems respond to drought. This paper aims to fill this gap by providing a global map of θ<sub>crit</sub> and identifying the factors that drive its spatial variation.
Methods
The core of this study lies in a novel approach to estimate θ<sub>crit</sub> using satellite data. The researchers leveraged the relationship between soil moisture (SM) and two key indicators of plant water stress:
- Evaporative Fraction (EF): The ratio of evapotranspiration to net radiation. When soil moisture is abundant (energy-limited regime), EF is high. As soil moisture decreases (water-limited regime), EF drops.
- Diurnal Amplitude of Land Surface Temperature (dLST): When plants are water-stressed, they reduce evapotranspiration, leading to more energy being released as sensible heat, which increases the temperature difference between day and night.
The study identified periods of "dry-downs" (extended periods without rain) from satellite observations. During these dry-downs, they analyzed the EF-SM and dLST-SM relationships. The point where dLST starts to increase (or EF starts to decrease) with decreasing SM was identified as θ<sub>crit</sub>. This method was validated using in-situ data from 44 flux tower sites. To understand the drivers of global θ<sub>crit</sub> variations, they employed explainable machine learning (random forest models) using 35 candidate environmental variables. Finally, they calculated the fraction of days per year when SM was below the estimated θ<sub>crit</sub> (Fraction of Stressed Days, FSD) and analyzed trends in FSD over the past four decades using reanalysis data.
Datasets
The study utilized several global datasets:
- Soil Moisture (SM) Datasets:
- SMAP-IB (version 1)
- SCA-V (version 7)
- SMOS-IC
- ERA5-Land reanalysis data (for trend analysis and model comparison)
- Land Surface Temperature (LST) Datasets:
- Copernicus Global Land Operations (hourly data)
- MODIS (Terra and Aqua) (four observations per day)
- In-situ Data: Eddy covariance measurements (SM, latent heat flux, sensible heat flux, outgoing longwave radiation) from 279 flux tower sites (ICOS, AmeriFlux, FLUXNET2015).
- Environmental Variables: A comprehensive set of 35 factors including soil properties, vegetation structure, plant hydraulic traits, and climatic variables.
- Land Cover Change Data: From Song, Hansen (100) to mask areas with significant land cover changes for trend analysis.
Key Insights and Ideas
- Global θ<sub>crit</sub> Map is Possible: The study demonstrates that combining satellite-derived dLST and SM during dry-downs is a robust method for creating a global map of θ<sub>crit</sub>, overcoming limitations of previous localized or model-based approaches.
- Aridity Drives θ<sub>crit</sub>: The most significant finding is that θ<sub>crit</sub> is strongly influenced by aridity. Arid ecosystems have lower θ<sub>crit</sub>, suggesting plants there are adapted to extract water more efficiently even at very low soil moisture levels. Conversely, humid ecosystems have higher θ<sub>crit</sub>, indicating they can sustain more leaf area without being limited by water.
- Plant Adaptation is Key: The observed global pattern of θ<sub>crit</sub> is not random but reflects evolutionary and physiological adaptations of plants to their local water availability and atmospheric demand.
- Earth System Models Struggle: Current Earth System Models (ESMs) tend to overestimate θ<sub>crit</sub> in dry regions and underestimate it in wet regions, highlighting a significant bias in how they represent plant water stress. This could lead to inaccurate projections of carbon uptake and ecosystem responses to drought.
- Increasing Water Stress: The frequency of days where ecosystems experience water stress (SM below θ<sub>crit</sub>) has significantly increased globally over the last 40 years, indicating a growing vulnerability of terrestrial ecosystems.
Key Results
- Average Global θ<sub>crit</sub>: 0.19 m³/m³.
- Range of θ<sub>crit</sub>: Varies from 0.12 m³/m³ in arid ecosystems to 0.26 m³/m³ in humid ecosystems.
- Top Drivers of θ<sub>crit</sub>: Aridity index (most influential), followed by leaf area index (LAI) and sand fraction of soil texture.
- Crop Type Differences: Irrigated rice showed significantly higher θ<sub>crit</sub> than maize, wheat, and potato.
- Trend in Stressed Days: The annual fraction of days with water stress has increased globally by approximately one day per year over the past four decades.
- ESM Biases: ESMs showed less spatial variability in θ<sub>crit</sub> than observed, overestimating it in dry areas and underestimating it in wet areas.
Limitations and Future Work
- Surface SM vs. Rootzone SM: The study primarily uses surface SM, acknowledging that plants are sensitive to rootzone moisture. However, they cite previous work suggesting surface SM can be a good proxy, and their analysis with ERA5-Land data supports this.
- Spatial Scale Mismatch: Acknowledged discrepancy between the footprint of flux tower measurements (meters to tens of meters) and satellite observations (25 km), which can contribute to differences in estimated θ<sub>crit</sub>.
- Temporal Dynamics of θ<sub>crit</sub>: While the study assumes θ<sub>crit</sub> is relatively stable over 40 years for trend analysis, they note that some research suggests temporal variations in θ<sub>crit</sub>, indicating a need for future research to better understand its dynamics.
- ESM Parameterization: The study highlights that ESMs need to improve the parameterization of plant hydraulic traits and stomatal closure responses to better represent water stress, especially in arid regions.
- Future Research Directions: The authors suggest further investigation into wet regions with few dry-downs, the temporal dynamics of θ<sub>crit</sub>, and the integration of spatially explicit θ<sub>crit</sub> parameterizations into land surface models to improve climate and water resource projections.
Summary updated