 |
| Titel |
Explaining the vertical soil organic matter distribution with lead-210 measurements and Bayesian calibration |
| VerfasserIn |
Maarten Braakhekke, Markus Reichstein, Thomas Wutzler, Christian Beer, Marcel Hoosbeek, Bart Kruijt, Jens Kattge, Marion Schrumpf, Pavel Kabat |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250054832
|
|
|
|
|
|
| Zusammenfassung |
| Since most conditions that control decomposition (e.g. soil organic matter (SOM) quantity
and quality, temperature and moisture, microbial dynamics) vary strongly with depth, the
vertical distribution of organic matter in the soil profile and surface organic layers plays an
important role for soil organic carbon cycling. Therefore, soil carbon models need to move
towards a more vertically explicit representation of the soil in order to be robust over different
soils and ecosystems, and under changing environmental conditions. Unfortunately, the
processes that determine the vertical distribution of SOM in the soil are still poorly
understood and quantified. In general three processes contribute to carbon deposition in the
profile: rhizodeposition, mixing due to bioturbation, and movement with the liquid
phase as dissolved organic matter. Due to the convolution of these processes, the
source of organic matter at a given depth cannot be determined simply from the
local concentration. Hence, the vertical SOM distribution alone is not sufficient to
parameterize a process-oriented model of SOM profile dynamics; additional information is
needed.
The profile of the radioactive isotope lead-210 may offer this information. 210P b is a
cosmogenic isotope, produced both in the soil and in the atmosphere. Deposition of
atmospheric 210P b causes surface enrichment in the soil in excess to the in situ produced
lead-210. The surface enrichment (210P bex ) is reduced by vertical transport and decay.
Since 210P b is input only at the soil surface and binds strongly to organic matter, 210Pbex
can be used as a tracer for soil organic matter transport.
We performed a Bayesian parameter estimation of the previously developed mechanistic
SOM profile model SOMPROF that includes the three above mentioned processes. Using a
Markov Chain Monte Carlo algorithm, 13 parameters, related to decomposition
and transport of organic matter, were calibrated for two sites: a coniferous forest
(Loobos, the Netherlands) on a poor sandy soil, and a deciduous forest with a rich
clay soil (Hainich, Germany). We used measurements of organic carbon stocks
and concentrations, and turnover rate measurements. Furthermore, we performed
a series of optimizations in which we stepwise added 210Pbex data and prior
knowledge of the parameters, to determine the value of the different sources of
information.
The results of the initial optimizations, in which 210Pbex and/or priors were
omitted, reveal the existence of several regions in the parameter space that lead to
reasonable fits to the measurements. These regions are characterized by the dominant
mechanism for soil carbon input: root litter production, bioturbation or liquid phase
transport. However, when additional information is included, one clearly favorite region
in parameter space can be identified for each site. This shows that the 210P bex
measurements as well as prior knowledge about the parameter values are very useful
for parameterizing the model, allowing the rates of the different processes to be
estimated.
The results of the final optimization suggest that for both sites advective transport
(movement as dissolved organic matter) is a dominant mechanism for organic matter input in
the mineral soil. However, for Hainich diffusion (bioturbation) plays a greater role
than for Loobos, which has much stronger advection. These results are in good
agreement with the differences between the two sites in terms of texture and biological
activity. |
|
|
|
|
|
|