 |
| Titel |
Stochastic Downscaling for Hydrodynamic and Ecological Modeling of Lakes |
| VerfasserIn |
D. Schlabing, M. Eder, M. Frassl, K. Rinke, A. Bárdossy |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250069743
|
|
|
|
|
|
| Zusammenfassung |
| Weather generators are of interest in climate impact studies, because they allow
different modi operandi: (1) More realizations of the past, (2) possible futures as
defined by the modeler and (3) possible futures according to the combination of
greenhouse gas emission scenarios and their Global Circulation Model (GCM)
consequences.
Climate modeling has huge inherently unquantifiable uncertainties, yet the results present
themselves as single point values without any measure of uncertainty. Given this reduction of
risk-relevant information, stochastic downscaling offers itself as a tool to recover the
variability present in local measurements. One should bear in mind that the lake models that
are fed with downscaling results are themselves deterministic and single runs may prove to be
misleading. Especially population dynamics simulated by ecological models are sensitive to
very particular events in the input data. A way to handle this sensitivity is to perform
Monte Carlo studies with varying meteorological driving forces using a weather
generator.
For these studies, the Vector-Autoregressive Weather generator (VG), which was first
presented at the EGU 2011, was developed further. VG generates daily air temperature,
humidity, long- and shortwave radiance and wind. Wind and shortwave radiation is
subsequently disaggregated to hourly values, because their short term variability has proven
important for the application. Changes relative to the long-term values are modeled as
disturbances that act during the autoregressive generation of the synthetic time series. The
method preserves the dependence structure between the variables, as changes in the disturbed
variable, say temperature, are propagated to the other variables. The approach is flexible
because the disturbances can be chosen freely. Changes in mean can be represented as
constant disturbance, changes in variability as episodes of certain length and amplitude. The
disturbances can also be extracted from GCMs with the help of QQ-downscaled time
series.
Results of water-quality and ecological modeling using data from VG is contributed by
Marieke Anna Frassl under the title “Simulating the effect of meteorological variability on a
lake ecosystem”. Maria Magdalena Eder contributes three dimensional hydrodynamic lake
simulations using VG data in a poster entitled “Advances in estimating the climate sensibility
of a large lake using scenario simulations”. Both posters can be found in the Session “Lakes
and Inland Seas” (HS10.1). |
|
|
|
|
|
|