Remote Sensing of Plant Functional Traits for Modeling Arctic Tundra Carbon Dynamics
| Active Dates | 8/15/2020-8/14/2024 |
|---|---|
| Program Area | Environmental Systems Science |
Project Description
Permafrost
carbon feedbacks to global climate are a major concern because the Arctic is warming twice as fast as other regions and >50% of the global belowground organic carbon (C) pool is stored in permafrost and overlying soils. Thawing of permafrost and frozen soils can expose organic C to decomposition, potentially converting these
ecosystems
into sources of
CO2
and accelerating warming. Plant communities in the Arctic
tundra
are also responding to warming, and these changes are expected to alter key biogeochemical and physical processes that feedback to climate. However, the direction and magnitude of effects are highly uncertain because tundra vegetation responses to warming are non-uniform and poorly represented in models.
The proposed research develops capacity for the remote estimation of aboveground and belowground plant functional traits and streamlines their inclusion in process models to quantify and predict regional C balance. Specifically, we will characterize directly observable plant functional traits from remotely sensed data; predict non-observable (e.g., belowground) traits by leveraging trait-environment relationships and trait covariation; and integrate this information into a newly developed modeling framework to quantify and predict terrestrial C stocks, fluxes, and their uncertainties in the Alaskan tundra. We will test four hypotheses: (1) Plant functional traits are predominantly shaped by climate, with local soil moisture moderating trait response to climate, but optimal trait values for a given environment depend on community type because plant functional types (PFTs) have different sensitivities to altered resources associated with climate warming. (2) Trait dispersion depends on environmental stress; where environmental conditions are more favorable for growth, there is greater variation in traits and overall greater trait diversity; (3) Given trait tradeoffs and trait-environment interactions, root traits are predictable from leaf and size traits, climate and soil factors; and (4) Variation in traits affect Arctic C balance. Leaf traits as they relate to photosynthetic parameters will have the strongest effects on C uptake. Root traits will exhibit fewer direct effects on C uptake but will be important in supporting nutrient uptake that then influences C uptake. To test these hypotheses, we will study the variation in plant traits across the major tundra vegetation communities present along local soil gradients nested within a macroclimate gradient in northern Alaska, quantifying trait-environment relationships and trait covariation at the community- and PFT-levels. Using machine-learning approaches, we will integrate ground-based measurements with information derived from multi-scale remote sensing platforms from drone, hypertemporal, LiDAR, and hyperspectral imagery to produce maps of leaf, size and root traits, greatly expanding the trait information available for modelers. We will use Bayesian data-model fusion methods to improve the parameterization and formulation of the Terrestrial Ecosystem Model, which is widely used in Arctic carbon studies, and perform simulation experiments to evaluate how differences in plant functional traits affect C dynamics. To synergize the trait-based modeling community, we will create a workshop to share findings and strategies for improving the representation of trait variation in ecosystem and Earth System Models (ESMs).
The proposed research directly supports the DOE near-term priorities by improving understanding of the interactions and feedbacks among above- and belowground plant system components and related soil-plant-atmosphere processes in a region that is inadequately represented in ESMs. Current models reduce the complexity of Arctic vegetation to a small number of PFTs. This approach implicitly assumes that each PFT represents the average ecological function of its constituent species, thus ignoring the effects of trait variation on carbon cycling and potentially leading to large uncertainty in the sign and magnitude of ecosystem feedbacks to climate. We will fill critical gaps in our understanding of the dominant PFTs and facilitate the development of alternative modeling approaches that allow the traits of PFTs to vary, thereby vastly enhancing the capacity of simulation models to project ecosystem carbon dynamics in a rapidly changing Arctic.
The proposed research develops capacity for the remote estimation of aboveground and belowground plant functional traits and streamlines their inclusion in process models to quantify and predict regional C balance. Specifically, we will characterize directly observable plant functional traits from remotely sensed data; predict non-observable (e.g., belowground) traits by leveraging trait-environment relationships and trait covariation; and integrate this information into a newly developed modeling framework to quantify and predict terrestrial C stocks, fluxes, and their uncertainties in the Alaskan tundra. We will test four hypotheses: (1) Plant functional traits are predominantly shaped by climate, with local soil moisture moderating trait response to climate, but optimal trait values for a given environment depend on community type because plant functional types (PFTs) have different sensitivities to altered resources associated with climate warming. (2) Trait dispersion depends on environmental stress; where environmental conditions are more favorable for growth, there is greater variation in traits and overall greater trait diversity; (3) Given trait tradeoffs and trait-environment interactions, root traits are predictable from leaf and size traits, climate and soil factors; and (4) Variation in traits affect Arctic C balance. Leaf traits as they relate to photosynthetic parameters will have the strongest effects on C uptake. Root traits will exhibit fewer direct effects on C uptake but will be important in supporting nutrient uptake that then influences C uptake. To test these hypotheses, we will study the variation in plant traits across the major tundra vegetation communities present along local soil gradients nested within a macroclimate gradient in northern Alaska, quantifying trait-environment relationships and trait covariation at the community- and PFT-levels. Using machine-learning approaches, we will integrate ground-based measurements with information derived from multi-scale remote sensing platforms from drone, hypertemporal, LiDAR, and hyperspectral imagery to produce maps of leaf, size and root traits, greatly expanding the trait information available for modelers. We will use Bayesian data-model fusion methods to improve the parameterization and formulation of the Terrestrial Ecosystem Model, which is widely used in Arctic carbon studies, and perform simulation experiments to evaluate how differences in plant functional traits affect C dynamics. To synergize the trait-based modeling community, we will create a workshop to share findings and strategies for improving the representation of trait variation in ecosystem and Earth System Models (ESMs).
The proposed research directly supports the DOE near-term priorities by improving understanding of the interactions and feedbacks among above- and belowground plant system components and related soil-plant-atmosphere processes in a region that is inadequately represented in ESMs. Current models reduce the complexity of Arctic vegetation to a small number of PFTs. This approach implicitly assumes that each PFT represents the average ecological function of its constituent species, thus ignoring the effects of trait variation on carbon cycling and potentially leading to large uncertainty in the sign and magnitude of ecosystem feedbacks to climate. We will fill critical gaps in our understanding of the dominant PFTs and facilitate the development of alternative modeling approaches that allow the traits of PFTs to vary, thereby vastly enhancing the capacity of simulation models to project ecosystem carbon dynamics in a rapidly changing Arctic.
Award Recipient(s)
- University of Illinois Urbana-Champaign (PI: Fraterrigo, Jennifer)