 |
| Titel |
Comparing measured and modelled soil carbon: which site-specific variables
are linked to high stability? |
| VerfasserIn |
Andy Robertson, Meagan Schipanski, Liwang Ma, Lajpat Ahuja, Niall McNamara, Pete Smith, Christian Davies |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250136987
|
| Publikation (Nr.) |
 EGU/EGU2016-18154.pdf |
|
|
|
|
|
| Zusammenfassung |
| Changes in soil carbon (C) stocks have been studied in depth over the last two decades, as net
greenhouse gas (GHG) sinks are highlighted to be a partial solution to the causes of climate
change. However, the stability of this soil C is often overlooked when measuring these
changes. Ultimately a net sequestration in soils is far less beneficial if labile C is replacing
more stable forms. To date there is no accepted framework for measuring soil C
stability, and as a result there is considerable uncertainty associated with the simulated
impacts of land management and land use change when using process-based systems
models. However, a recent effort to equate measurable soil C fractions to model
pools has generated data that help to assess the impacts of land management, and
can ultimately help to reduce the uncertainty of model predictions. Our research
compiles this existing fractionation data along with site metadata to create a simplistic
statistical model able to quantify the relative importance of different site-specific
conditions.
Data was mined from 23 published studies and combined with original data to generate a
dataset of 100+ land use change sites across Europe. For sites to be included they required
soil C fractions isolated using the Zimmermann et al. (2007) method and specific site
metadata (mean annual precipitation, MAP; mean annual temperature, MAT; soil pH; land
use; altitude). Of the sites, 75% were used to develop a generalized linear mixed
model (GLMM) to create coefficients where site parameters can be used to predict
influence on the measured soil fraction C stocks. The remaining 25% of sites were
used to evaluate uncertainty and validate this empirical model. Further, four of the
aforementioned sites were used to simulate soil C dynamics using the RothC, DayCent
and RZWQM2 models. A sensitivity analysis (4096 model runs for each variable
applying Latin hypercube random sampling techniques) was then used to observe
whether these models place as much weight on the same site parameters as the
GLMM.
Sites were spread across an extensive geographic area and encompassed a wide range of
conditions (2% to 44% clay content; 0.9˚ C to 18˚ C MAT; 300mm to 1400mm MAP).
Topsoil (30 cm) C stocks also varied considerably (29.0 to 115.9 t/ha) but the proportion
deemed stable (mean residence time >10 years) was relatively consistent (72 ± 2 %). The
GLMM approach suggested that an interaction of soil pH and historic land use explained the
largest amount of variation seen in stable fraction C stocks, closely followed by
MAT and MAP interactions. For all three systems models, the stable soil C pools
were most sensitive to climatic variables and land use. However, RZWQM2 did
indicate that soil characteristics (texture, pH) also had an influence on stable C pool
dynamics.
References
1 – Zimmermann et al., 2007. Measured soil organic matter fractions can be
related to pools in the RothC model. European Journal of Soil Science, 58:658–667. |
|
|
|
|
|
|