 |
| Titel |
Using a data-assimilation system to assess the influence of fire on simulated carbon fluxes and plant traits for the Australian continent |
| VerfasserIn |
Jean-François Exbrayat, T. Luke Smallman, A. Anthony Bloom, Mathew Williams |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250106741
|
| Publikation (Nr.) |
 EGU/EGU2015-6421.pdf |
|
|
|
|
|
| Zusammenfassung |
| Natural disturbances, such as fire, play an important role in the carbon balance of terrestrial
ecosystems. Both burned emissions and the impact of fire on plant growth must be
considered to quantify the magnitude of the current and future terrestrial carbon
sink. However, fire is rarely represented in Earth System Models, and the usual
classification of ecosystems in a limited number of global plant functional types
does not take into account local adaptations to fire regimes that enable resilience of
ecosystems.
We show the importance of these mechanisms with a terrestrial model-data fusion scheme
applied to the fire-prone Australian continent. We use the CARbon DAta-MOdel fraMework
(CARDAMOM) to assimilate time series of MODIS LAI and GFED burned area and
use the Harmonized World Soil Database and remote-sensing based estimates of
Above-Ground Biomass as prior knowledge for initial conditions. In each pixel, a
Markov Chain Monte-Carlo algorithm is used to optimise parameters according to
observations. Meanwhile, ecological and dynamical constraints representative of
real world processes constrain parameter inter-dependencies and long-term pool
dynamics. CARDAMOM outputs maps of ecosystem carbon fluxes and parameters as
well as their uncertainty sampled from the posterior distribution provided by the
MCMC.
We perform two data-assimilations over Australia. The first experiment is a control run
that includes fire drivers while the second experiment does not consider the occurrence of
fires. Results of the first experiment are comparable to previous estimates and show that
Australian ecosystems have most likely been acting as a carbon sink since the year 2000 with
a large fire-driven inter-annual variability (best estimate of 264 ± 172 Tg C yr-1).
However, our results indicate that the most intense fire seasons may temporarily turn the
continent into a net source of carbon offsetting the natural carbon sink of the same
year.
Comparing the parameter maps generated with and without fire clearly indicates that
frequently burned ecosystems tend to optimise their net carbon uptake through different
means (higher NPP:GPP ratio, higher canopy efficiency, etc/¦) to cope with the repeated
removal of above ground biomass. Our parameter maps, which are comparable to plant traits
maps, show that the plant functional type concept currently used in Earth System Models is
likely not adapted for ecosystems that experience frequent disturbances. Considering
ecosystem disturbances and going beyond the plant functional type concept are two urgent
requirements to improve terrestrial carbon models and projections of the terrestrial
carbon-climate feedback. |
|
|
|
|
|
|