|
Titel |
Upscaling of annual mean and dynamics of water table depth in German organic soils |
VerfasserIn |
Michel Bechtold, Bärbel Tiemeyer, Susanne Belting, Andreas Laggner, Thomas Leppelt, Enrico Frahm, Annette Freibauer |
Konferenz |
EGU General Assembly 2013
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 15 (2013) |
Datensatznummer |
250082004
|
|
|
|
Zusammenfassung |
Water table depth is the key parameter controlling the fluxes of CO2, CH4 and N2O from
organic soils (peatlands and other organic soils). Therefore, a good estimation of the
spatial distribution of water table depth is crucial in any upscaling approach for
these greenhouse gases (GHGs). It is further the prerequisite to assess the effects
of re-wetting measures. There are attempts to obtain maps of water table depth
at large scales (e.g. national or continental) by using process-based hydrological
model concepts. However, major problem of the process-based approach is the
representation of the water management (ditches, tile drains, pumping and weir
management), which is at the best known spatially just for the ditch patterns. Thus, this
approach is hardly applicable to the diversely-drained and -used organic soils in central
Europe.
Here, we present an alternative, data-driven approach for upscaling annual mean and
dynamics of water table depth in organic soils. Groundwater level data of a unique dataset
from about 60 peatlands, 1100 dipwells and around 8000 annual data sets, is the
basis of this approach. Time series were used to calculate long-term annual means,
average annual amplitudes and ponding durations. In case of continuous observations,
shape parameters of the annual frequency distribution of water table depths were
calculated. For each well, numerous site characteristics were collected as possible
explanatory variables. This collection was restricted to nationally-available data. For each
dipwell, land use is taken from official land use maps (German database ATKIS), and
the soil type from the national geological map (1:200.000). In case of reliable site
information, maps were corrected accordingly. Additionally, from these maps, topological
indicators such as the ditch distance and density, the distance to the edge of the
peatland and the peatland area within different buffers were calculated. Meteorological
data (precipitation, potential evapotranspiration and climatic water balance) was
extracted from gridded data (1x1 km) from the German Weather Service. Topographic
indices were calculated using the national digital elevation model. Further, protection
status (nature reserves, Natura2000, etc.) and peatland type was collected for each
well.
We use two data-driven models (fuzzy-logic and boosted regression trees) to analyze the
influence of the site characteristics on the various water table depth target variables
(mean, amplitude, etc.). First results using the fuzzy-logic approach show that a
land use/vegetation and protection status categorization of the data combined with
separate fuzzy models for each category can explain substantial parts of the variance
seen in the data set. Variables with strong explaining power were meteorological
(summer precipitation and/or climatic water balance) and topological parameters of
the ditch network and the peatland body. Uncertainty of the models is evaluated
using cross-validation. Models are applied with nationally-available data to generate
maps of statistical measures of water table depth for the German organic soils. |
|
|
|
|
|