Research Highlights

With satellite remote sensing observations, my main research questions are: how does the land surface respond to climate variability and extremes? Which soil-plant mechanisms drive this land surface response?

Answering this question typically leads to deeper investigations of dryland ecosystems, which cover ~40% of the globe and are a major driver of carbon cycle interannual variability. I also am investigating use of different remote sensing tools to increase our capability to observe the water and carbon cycles of the terrestrial biosphere. Ultimately, observing landscape-scale relationships of water-carbon cycle coupling in the biosphere informs global modeling efforts.

My research projects fall into three related categories:

1) Dryland Vegetation: response to rain pulses, rain frequency/intensity changes, and extremes
a) How quickly does dryland vegetation take up rain pulse water compared to more mesic environments?

Figure: Global plant water uptake timescales from SMAP microwave remote sensing
Summary: Dryland vegetation takes a few days to increase its plant water content due either to slow rehydration under initial dry conditions or due to rapid biomass growth after larger pulses. Flux tower measurements also show gradual carbon flux increases after storms in drylands. By contrast, vegetation in mesic environments quickly equilibrates with soil moisture with short-lived carbon uptake responses after a rain pulse.
Reference: See studies in Biogeosciences and Water Resources Research 2021 studies.

b) How do land-atmosphere factors drive losses in plant saturation?
Summary: Based on a Granger causality model, soil moisture loss is a strong driver of plant water loss, while post-storm temperature and vapor pressure deficit increases also further drive plant water loss regionally. During drying, soil moisture and temperature interact to further reinforce plant drying, providing mechanistic insight into land surface evolution during initial stages of plant water stress and drought. These effects occur mainly in the water-limited regime and are strongest and most prevalent in drier environments.
Reference: See Geophysical Research Letters 2020 study.

c) Is there widespread evidence for the “pulse reserve hypothesis” where plants quickly invest in biomass growth after biologically significant rain pulses?

Figure: Strength of plant response to moisture pulses from SMAP microwave remote sensing
Summary: New vegetation optical depth observations from the SMAP satellite show that tropical vegetation frequently shows strong plant water uptake and growth responses to individual rain events, especially those that are strong enough to pulse moisture above soil moisture thresholds. As such, the "pulse-reserve" paradigm proposed for dryland vegetation is more widespread than previously thought occurring in at least 50% of tropical vegetated surfaces. Pulse-reserve behavior is indeed most widespread and strongest in drylands, but is still present with progressively decreasing strength in wetter environments.
Reference: See Nature Plants 2018 study.

d) How are plants responding to global changes in rainfall frequency and intensity?
Summary: research ongoing

e) How are plants in climate change hotspots like the West US responding to extreme drought and heatwave events?

Summary: research under peer review
 

 

 

2) Land Surface Hydrology: landscape response to forcing
a) Where do land surfaces respond most to climatic perturbations and why? 

Figure: Land surface responsiveness based on diurnal land surface temperature variations from SEVIRI
Summary: Locations that spend more time in a "water-limited regime" (where soil moisture variations strongly influence evaporation and sensible heat flux) are most responsive to forcing. We found that these tend to be semi-arid regions. Semi-arid regions therefore have the most vulnerability to interannual variations and trends in rainfall and solar radiation. This study was conducted solely with observations to inform modeling efforts.
Reference: See Water Resources Research 2022 study.

 

b) Can we use remote sensing to observe landscape water-limitation and energy fluxes responses to soil moisture?

Figure: (Left) Energy flux-soil moisture relationships at FLUXNET site. (Right) Diurnal temperature information from remote sensing can be used to observe (without models) water and energy limitation at a location
Summary: We find the diurnal surface temperature amplitude (dT) reflects surface energy flux information, inversely related to evaporation. Given that large-scale evaporation datasets require models and model assumptions, diurnal temperature satellite retreivals along with satellite soil moisture retrievals allow direct observation of evaporation-soil moisture relationships. With these relationships, we determined the observed degree of land-atmosphere coupling, soil moisture thresholds, and time-spent in the water-limited evaporative regime across Africa.
Reference: See Water Resources Research 2019 study.

 

c) How does vegetation cover influence the surface energy balance, and specifically land surface temperature, across the tropics?

Figure: Interannual effect of vegetation (MODIS NDVI) on LST determined in units of estimated change in LST per 1% increase in mean NDVI (stippling p<0.05). (Bottom left) Distribution of values in drylands, humid regions, and all together. (Bottom right) Percent reduction in biophysical cooling due to drylands (all regions p<0.05). 

Summary: Plants net cool the surface across the tropics, but drylands’ magnitude of cooling is strongly (~50%) reduced due to their (1) reduced ability to cool with lower soil moisture and (2) tendency to increase solar radiation absorption. Cooling feedbacks with greening are weaker than expected in drylands. Caution should be shown for planned tree planting efforts in dryland locations given vegetation’s reduced ability to cool in these locations.

Reference: See Global Change Biology 2023 study.

 

3) Remote Sensing: exploring cutting-edge observing capabilities for the terrestrial biosphere

3.1 Microwave Remote Sensing

a) Can we retrieve soil moisture and plant water content from microwave L-band satellites? Can we observe these properties in forests?


Figure: Microwave radiative transfer considerations for forested landscapes
Summary: Here, we formulated a simple first-order scattering model that is able to characterize multiple-scattering of microwaves from mainly woody biomass that is neglected in traditional soil moisture retreival algorithms. We are able to partition out and detect first-order scattering contribution to the microwave satellite signal, which increases the sensitivity to the surface soil moisture. 

Reference: See Remote Sensing of Environment 2018 study.

b) Do L-band microwave satellites like SMAP only observe soil moisture in the top 0-5 cm soil layers?

Summary: Evidence from field and statistical experiments show satellite (L-band) soil moisture representation often extends below 5 cm because (1) dry soil microwave emission origins from deeper layers and (2) wet soils have strong vertical correlation. 5 cm is not a "cutoff" of representation. Additionally, isotopic tracer studies show many cases of shallow plant water use making satellite L-band soil moisture application useful for many grassland, shrubland, and lightly-wooded savanna applications.

Reference: See Water Resources Research 2023 commentary.

c) How do satellite instrument errors propagate into soil moisture and vegetation optical depth retrievals and can we reduce these errors?

Figure: Cost function curvature for joint retrievals of soil moisture and VOD
Summary: We first estimate errors in satellite soil moisture and VOD retrievals in their joint retrieval using simultaneous retrieval algorithms (dual channel algorithm). We find that errors are largest in denser woody biomass regions and that random satellite measurement error nearly always propagates more into VOD than soil moisture. However, use of "regularization" reduces the error mainly in VOD as well as reduces spurious compensation of errors between soil moisture and VOD in their retrieval. Regularization (as with the SMAP MTDCA and a newer SMAP MDCA algorithm) allows interpretation of VOD temporal variability at shorter sub-seasonal timescales.

Reference: See IEEE JSTARS 2021 study

3.2 Remote Sensing of Atmospheric Carbon Concentrations

Can we use XCO2 column retrievals as an early detection to rapidly investigate surface carbon flux anomalies in the biosphere?

Figure: In many global locations, correlation of monthly OCO-2 column carbon dioxide concentration anomalies consistently increasing when the terrestrial biosphere carbon uptake decreases despite strong role of atmospheric transport.
Summary: OCO-2 column carbon dioxide retrievals, available at 1–3 month latency, can be used to directly detect and roughly estimate extreme biospheric COfluxes. CarbonTracker reanalysis is used to establish domain area and wind conditions required for the methods. A caution that both atmospheric transport and OCO-2 instrument noise limit the application to specific regions, times of year, and only to extreme anomalies.

Reference: See Atmospheric Chemistry and Physics 2023 study.