Global Respiration Comparison (ResCom): evaluation of satellite constrained top-down and bottom-up respiration estimates and their relationship with model simulations

Ashley Ballantyne,  University of Montana,  ashley.ballantyne@umontana.edu (Presenter)

One of the greatest sources of uncertainty for future climate predictions is the strength of
climate feedbacks due to respiration, the most uncertain process in the global carbon
cycle. Our understanding of global respiration has been limited by observations at the
global scale that have restricted model testing and development. However, a proliferation
of soil flux measurements now make it possible to resolve bottom-up estimates of
respiration from regional to global scales, while extensive atmospheric measurements
combined with recent satellite observations enable top-down estimates at global to
regional scales. This massive increase in observations has been accompanied by the
rapid development and application of machine learning (ML) approaches in the Earth
sciences. It is also possible to partition bottom-up and top-down respiration into its
component fluxes of autotrophic (Ra) and heterotrophic respiration (Rh) at regional and
decadal timescales. This is important because Rh ,is highly sensitive to environmental
drivers that are expected to cause enhanced respiratory carbon losses in the future.
Recent model developments and reporting allow for Ra and Rh to be further partitioned
between aboveground and belowground fluxes, allowing for direct comparison with our
bottom-up and top-down respiration estimates.
The overall objective of the ResCom project is to improve our understanding of global
respiration by combining spaceborne observations with extensive surface measurements
in a ML framework to provide spatio-temporally continuous estimates of bottom-up and
top-down respiration for benchmarking models. Specifically, we will address the
following research questions: 1. What are the current mean annual estimates of bottomup and top-down Rh and how well do these estimates compare with each other? 2. Can
the spatial and temporal resolutions of our bottom-up and top-down estimates be
improved by additional in situ measurements, remote sensing observations, and novel
machine learning approaches to reveal important temporal and regional mismatches
attributable to carbon cycle processes? and 3. How well do our newly derived spatially
and temporally explicit estimates of bottom-up and top-down Rh compare with historical
and future model simulations of Rh?
To answer these research questions we will relate continuous and soil respiration
measurements to a suite of remotely sensed environmental variables in our ML
framework to derive spatially continuous monthly estimates of bottom-up Rh. At the
same time, we will use a range of remotely sensed estimates of primary productivity and
net biome exchange of CO2, in combination with an empirically derived estimate of
carbon use efficiency, to derive spatially continuous monthly estimates of top-down Rh.
Both of the bottom-up and top-down Rh data products will be generated from 1990 to
2020 with fully characterized uncertainties. These bottom-up and top-down Rh estimates
will be compared first with each other to identify significant differences in mean annual
estimates as well as mismatches at seasonal and regional scales, and then compared to
historical simulations from dynamic vegetation and Earth system models. Lastly, we will
use our optimized ML estimates of Rh as an emergent constraint on future simulations of
carbon fluxes. The ResCom project is made possible by NASA remote sensing products,
and its goals fully align with NASA research priorities.

Presentation: Talk_Ballantyne_209_35.pptx 

Presentation Type: Talk

Session: TE Plenary Morning Session 1

Talk Time: Tue (May 9) 9:25 AM