Estimating global ecosystem iso/anisohydry using active and passive microwave satellite data


Yan Li\affil1, Kaiyu Guan\affil1,2, Pierre Gentine\affil3,4, Alexandra G. Konings\affil5, Frederick C. Meinzer\affil6, John S. Kimball\affil7, Xiangtao Xu\affil8, William R. L. Anderegg\affil9, Nate G. McDowell\affil10, Jordi Martínez-Vilalta\affil11,12, David G. Long\affil13, Stephen P. Good\affil14


1Department of Natural Resources and Environmental Sciences, University of Illinois at Urbana–Champaign, IL 61801, USA \affiliation2National Center for Supercomputing Applications, University of Illinois at Urbana–Champaign, IL 61801, USA \affiliation3Department of Earth and Environmental Engineering, Columbia University, New York, NY 10027, USA \affiliation4Earth Institute, Columbia University, New York, NY 10027, USA \affiliation5Department of Earth System Science, Stanford University, Stanford, CA 94305, USA \affiliation6USDA Forest Service, Pacific Northwest Research Station, Corvallis, OR 97331 USA \affiliation7Numerical Terradynamic Simulation Group, College of Forestry & Conservation, University of Montana, Missoula, MT 59812, USA \affiliation8Department of Geosciences, Princeton University, Princeton, NJ 08544, USA \affiliation9Department of Biology, University of Utah, Salt Lake City, UT 84112, USA \affiliation10Pacific Northwest National Laboratory, Richland, WA 99354, USA \affiliation11CREAF, Cerdanyola del Vallès E-08193 (Barcelona), Spain \affiliation12Univ. Autònoma Barcelona, Cerdanyola del Vallès E-08193 (Barcelona), Spain \affiliation13Department of Electrical and Computer Engineering, Brigham Young University, Provo, Utah 84602, USA \affiliation14Department of Biological and Ecological Engineering, Oregon State University, Corvallis, Oregon 97330, USA


Yan \correspondingauthorKaiyu \journalnameJGR-Biogeosciences


The concept of iso/anisohydry describes the degree to which plants regulate their water status, operating from isohydric with strict regulation to anisohydric with less regulation. Though some species-level measures of iso/anisohydry exist at a few locations, ecosystem-scale information is still largely unavailable. In this study, we use diurnal observations from active (Ku-Band backscatter from QuikSCAT) and passive (X-band Vegetation Optical Depth [VOD] from AMSR-E) microwave satellite data to estimate global ecosystem iso/anisohydry. Here, diurnal observations from both satellites approximate predawn and midday plant canopy water contents, which are used to estimate iso/anisohydry. The two independent estimates from radar backscatter and VOD show reasonable agreement at low and mid-latitudes but diverge at high latitudes. Grasslands, croplands, wetlands, and open shrublands are more anisohydric, whereas evergreen broadleaf and deciduous broadleaf forests are more isohydric. The direct validation with upscaled in-situ species iso/anisohydry estimates indicates that the VOD-based estimates have much better agreement than the backscatter-based estimates. The indirect validation with prior knowledge suggests that both estimates are generally consistent in that vegetation water status of anisohydric ecosystems more closely tracks environmental fluctuations of water availability and demand than their isohydric counterparts. However, uncertainties still exist in the iso/anisohydry estimate, primarily arising from the remote sensing data and, to a lesser extent, from the methodology. The comprehensive assessment in this study can help us better understand the robustness, limitation, and uncertainties of the satellite-derived iso/anisohydry estimates. The ecosystem iso/anisohydry has the potential to reveal new insights into spatio-temporal ecosystem response to droughts.

Key words: isohydry/anisohydry, radar backscatter, vegetation optical depth, QuikSCAT, AMSR-E
Key points:

  • Iso/anisohydry estimates from backscatter and VOD data show reasonable agreement at low and mid-latitudes but diverge at high latitudes.

  • Grasslands, croplands, and open shrublands are more anisohydric while evergreen broadleaf and deciduous broadleaf forests are more isohydric.

  • VOD-based iso/anisohydry estimates show better agreement with upscaled in-situ measurements than backscatter-based estimates.


Plants have evolved to develop a range of ecophysiological strategies to withstand drought (Akinci et al., 2012; Farooq et al., 2012; Matheny et al., 2016). Among others, stomatal closure is a central physiological strategy by which plants limit their transpirational water loss in order to avoid hydraulic failure at short timescales (Martínez-Vilalta et al., 2016). The stringency of stomatal control differs among species and under various environmental conditions (Wong et al., 1979; Ball et al., 1987; Konings et al., 2017; Martínez-Vilalta et al., 2014), and co-varies with other plant hydraulic traits such as rooting depth (Martínez-Vilalta et al., 2016; Matheny et al., 2016). The resulting varying levels of regulation of plant water status can be characterized by the concept of iso/anisohydry — the continuum from isohydric to anisohydric behavior (Tardieu et al., 1998; McDowell et al., 2008; Martínez-Vilalta et al., 2014; Garcia-Forner et al., 2015; Landsberg et al., 2016; Meinzer et al., 2016; Martínez-Vilalta et al., 2016). Broadly speaking, isohydric plants can maintain a relatively stable midday leaf water potential as environmental conditions (e.g., atmospheric water demand and soil water content) change, thereby dampening the diurnal and seasonal variations in leaf water potential (Martínez-Vilalta et al., 2016; Meinzer et al., 2016). By contrast, anisohydric plants allow their leaf water potentials to closely track the fluctuations of environmental water availability and demand, and thus exhibit greater diurnal and seasonal variations. The range of behavior from isohydric to anisohydric is closely linked to stomatal responses to water deficits (Sperry et al., 2016) and has important implications for understanding plant response to different drought conditions (Roman et al., 2015; Konings et al., 2017a).

Modeling water-vegetation interactions still presents a major challenge, particularly plants’ responses to drought (McDowell et al., 2013). This is largely due to our knowledge gap in understanding these complex plant physiological behaviors at multiple scales (Fatichi et al., 2015; Landsberg et al., 2016). In current process-based terrestrial biosphere models, most representations of hydraulic processes are still simplistic and empirically-based (Fatichi et al., 2015), limiting their ability to capture and predict plant responses to water stress as well as their feedbacks to climate (Xu et al., 2013). There has been increasing recognition of the need for incorporating plant hydraulic processes into mechanistic models to improve the representation of fundamental aspects such as plant water stress and recovery, drought sensitivity, and ultimately mortality (Bonan et al., 2014; Fatichi et al., 2015; Anderegg et al., 2016; Xu et al., 2016; Konings et al., 2017a). Understanding the role of iso/anisohydry, which is an integrated consequence of stomatal control, hydraulic traits, and their interactions with the environment, could provide new insights to help improve the modeling of key (physiological) processes related to plant water regulation.

Contemporary understanding of iso/anisohydry and attempts to quantify it primarily come from studies conducted at the plant or species levels (Tardieu et al., 1998; Klein, 2014; Martínez-Vilalta et al., 2014), whereas much less is known about the variation of iso/anisohydry at ecosystem to global scales. At the ecosystem-scale, multiple species representing a range of isohydry to anisohydry co-exist, but large-scale gradients (e.g., regional or global scales) in the overall ecosystem behavior may still occur. Ecosystem-level iso/anisohydry information is more relevant to earth system modeling (e.g., for model benchmarking and parameterization) than species-level information, but it is not possible to obtain iso/anisohydry estimates at these large scales using traditional plant physiological methods. Spaceborne microwave remote sensing offers an unique opportunity. Both active and passive microwave remote sensing data are sensitive to vegetation water status, allowing them to be potentially useful for estimating ecosystem iso/anisohydry. Their sensitivities and ability to capture vegetation water status depend on operating frequency, overpass times, and physical properties of the land surface (Jones et al., 2011; Konings et al., 2017b; van Emmerik et al., 2015). Existing spaceborne records of both active and passive microwave have different characteristics on this front (Paget et al., 2016; Podest et al., 2014; Steele-Dunne et al., 2017). Recently, Konings et al. (2017) adapted the iso/anisohydry metric proposed by Martínez-Vilalta et al. (2014) to work with passive microwave satellite data from ASMR-E, and successfully produced the first global ecosystem-scale iso/anisohydry estimates. However, the reliability of such estimate from a single satellite record is uncertain due to potential sensor biases and retrieval errors. The degree to which these errors and varying sensitivities of microwave observation to vegetation water status affect the iso/anisohydry metric is unknown. Furthermore, a direct validation with in-situ measurements is lacking.

This study improves upon recent work (Konings et al., 2017) by addressing the aforementioned issues. First, we utilize both active and passive (with different retrieval algorithms) microwave satellite remote sensing data to provide new independent global iso/anisohydry estimates. Second, we validate these satellite-derived iso/anisohydry estimates through a cross-comparison between them and a direct comparison with upscaled in-situ species-level estimates. Third, we perform an indirect validation by examining whether the observed response of plant water status to environmental fluctuations follows the expected theoretical pattern of iso/anisohydry. The overall aim of this study is to promote better understanding of satellite-derived ecosystem iso/anisohydry estimates and the robustness, limitations, and uncertainties of these estimates, their potential applications and paths forward.