 |
| Titel |
Field Significance of Performance Measures in the Context of Regional Climate Model Verification |
| VerfasserIn |
Martin Ivanov, Kirsten Warrach-Sagi, Volker Wulfmeyer |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250103055
|
| Publikation (Nr.) |
 EGU/EGU2015-2456.pdf |
|
|
|
|
|
| Zusammenfassung |
| The purpose of this study is to rigorously evaluate the skill of dynamically downscaled global
climate simulations. We investigate a dynamical downscaling of the ERA-Interim
reanalysis using the Weather Research and Forecasting (WRF) model, coupled
with the NOAH land surface model within the scope of EURO-CORDEX. WRF
has a horizontal resolution of 11° and contains the following physics: the Yonsei
university atmospheric boundary layer parameterization, the Morrison two-moment
microphysics, the Kain-Fritsch-Eta convection and the Community Atmosphere Model
radiation schemes. Daily precipitation is verified over Germany for summer and winter
against high-resolution observation data from the German weather service for the first
time.
The ability of WRF to reproduce the statistical distribution of daily precipitation is
evaluated using metrics based on distribution characteristics. Skill against the large-scale
ERA-Interim data gives insight into the potential, additional skill of dynamical downscaling.
To quantify it, we transform the absolute performance measures to relative skill measures
against ERA-Interim. Their field significance is rigorously estimated and locally significant
regions are highlighted.
Statistical distributions are better reproduced in summer than in winter. In both seasons
WRF is too dry over mountain tops due to underestimated and too rare high and
underestimated and too frequent small precipitations. In winter WRF is too wet at
windward sides and land-sea transition regions due to too frequent weak and moderate
precipitation events. In summer it is too dry over land-sea transition regions due to
underestimated small and too rare moderate precipitations, and too wet in some river valleys
due to too frequent high precipitations. Additional skill relative to ERA-Interim is
documented for overall measures as well as measures regarding the spread and tails
of the statistical distribution, but not regarding mean seasonal precipitation. The
added value is spatially more widespread and regards more performance measures in
summer, while in winter it is concentrated in mountains. Precipitation intensity
is generally improved, deterioration is only documented in summer in land-sea
transition regions due to the underestimated small intensities. The frequencies of all
precipitation categories are improved in summer. In winter the moderate category shows no
skill; significant deterioration is observable due to overprediction of wet-days on
windward sides, of small precipitation events in the northern lowlands and of high
precipitation events in the North-West coast and the heaths north of the Ore mountains as
well as underprediction of high precipitation events in some concave topography
forms.
This study demonstrates in a rigorous manner the clear additional value of dynamical
downscaling over global climate simulations. |
|
|
|
|
|
|