 |
| Titel |
Temperature response functions introduce high uncertainty in modelled carbon stocks in cold temperature regimes |
| VerfasserIn |
H. Portner, A. Wolf, H. Bugmann |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250027794
|
|
|
|
|
|
| Zusammenfassung |
| Many biogeochemical models have been applied to study the response of the carbon cycle to
changes in climate, whereby the process of carbon uptake (photosynthesis) has usually gained
more attention than the equally important process of carbon release by respiration. The
decomposition of soil organic matter is driven by a combination of factors with a prominent
one being soil temperature [Berg and Laskowski(2005)]. One uncertainty concerns the
response function used to describe the sensitivity of soil organic matter decomposition
to temperature. This relationship is often described by one out of a set of similar
exponential functions, but it has not been investigated how uncertainties in the choice
of the response function influence the long term predictions of biogeochemical
models.
We built upon the well-established LPJ-GUESS model [Smith et al.(2001)]. We tested
five candidate functions and calibrated them against eight datasets from different Ameriflux
and CarboEuropeIP sites [Hibbard et al.(2006)]. We used a simple Exponential
function with a constant Q10, the Arrhenius function, the Gaussian function [Tuomi
et al.(2008), O’Connell(1990)], the Van’t Hoff function [Van’t Hoff(1901)] and
the Lloyd&Taylor function [Lloyd and Taylor(1994)]. We assessed the impact of
uncertainty in model formulation of temperature response on estimates of present and
future long-term carbon storage in ecosystems and hence on the CO2 feedback
potential to the atmosphere. We specifically investigated the relative importance of
model formulation and the error introduced by using different data sets for the
parameterization.
Our results suggested that the Exponential and Arrhenius functions are inappropriate, as
they overestimated the respiration rates at lower temperatures. The Gaussian, Van’t Hoff and
Lloyd&Taylor functions all fit the observed data better, whereby the functions of Gaussian
and Van’t Hoff underestimated the response at higher temperatures. We suggest, that the
function of Lloyd&Taylor therefore is an adequate choice to model the temperature
dependency of soil organic matter decomposition. The Ticino catchment (300-2300m) in
Southern Switzerland was used to study the sensitivity of long-term changes (100 years) in
the prediction of carbon storage. The uncertainty in temperature response introduced into
the model lead to high uncertainties in long-term soil carbon stocks. Interestingly,
the uncertainty increased with decreasing temperature and increasing elevation.
The carbon pools in lower elevations (mean annual temperature > 15-C) turned
over faster and little carbon accumulated in the soil. The carbon pools in higher
elevations and hence in higher latitudes experiencing colder temperature (mean annual
temperature < 15-C) turned over slower and therefore accumulated more carbon over the
simulation period. Therefore, the high elevation soils stored more carbon, but the
prediction of the carbon pool size had a much higher uncertainty than the low elevation
soils.
We concluded that with our model, the predictions of the potential loss of soil carbon in
cold temperature regimes is more uncertain than the carbon loss in warmer regions, both due
to the higher soil carbon pools, but also due to the higher uncertainty found in our
simulations. |
|
|
|
|
|
|