|
Titel |
Ensemble Postprocessing of Vertical Temperature Profiles using Standardized Anomalies |
VerfasserIn |
Thorsten Simon, David Schönach, Georg J. Mayr |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250144997
|
Publikation (Nr.) |
EGU/EGU2017-8887.pdf |
|
|
|
Zusammenfassung |
Statistical postprocessing of output from numerical weather prediction models has long
been used for bias-removal or further downscaling—mostly for single locations or
(quasi-)horizontal fields. This study ventures into the postprocessing of forecasts of vertical
temperature profiles. The shape of these profiles is a decisive factor for the occurrence of
thunderstorms and lightning.
The standardized anomalies model output statistics (SAMOS) approach is applied to
vertical temperature profiles observed with radiosondes. Basic idea of SAMOS is
summarized as follows: Firstly, climatologies are estimated from the observations and the
numerical weather prediction (NWP) output, respectively, in order to scale both data sets.
Secondly, the statistical model is built on the scaled data.
Observations are taken from stations in Germany (Bergen, Lindenberg, Idar-Oberstein,
Kümmersbruck) which launch radiosonde four times a day. The ECMWF-EPS
provides the prediction data. It contains 50 perturbed members and 1 control member.
ECMWF-reforecast data are used to estimate the model climatology and to train the statistical
model.
The climatologies are estimated with generalized additive models (GAM). They reveal
different mean vertical profiles for different launching hours, e.g., near-surface inversions are
visible at 00 UTC and 06 UTC. Moreover, annual cycles are larger closer to the surface.
Scaling reduces these features. Results and verification for the period of time from September
2016 to January 2017 will be shown. |
|
|
|
|
|