Articles | Volume 8, issue 1
Geosci. Model Dev., 8, 87–97, 2015
https://doi.org/10.5194/gmd-8-87-2015
Geosci. Model Dev., 8, 87–97, 2015
https://doi.org/10.5194/gmd-8-87-2015

Methods for assessment of models 23 Jan 2015

Methods for assessment of models | 23 Jan 2015

The impact of periodization methods on the kinetic energy spectra for limited-area numerical weather prediction models

V. Blažica1,*,**, N. Gustafsson2, and N. Žagar1 V. Blažica et al.
  • 1Department of Physics, Faculty of Mathematics and Physics, University of Ljubljana, Ljubljana, Slovenia
  • 2Swedish Meteorological and Hydrological Institute, Norrköping, Sweden
  • *now at: Slovenian Environment Agency, Ljubljana, Slovenia
  • **Invited contribution by V. Blažica, recipient of the EGU Outstanding Student Poster Award 2013.

Abstract. The paper deals with the comparison of the most common periodization methods used to obtain spectral fields of limited-area models for numerical weather prediction. The focus is on the impact that the methods have on the spectra of the fields, which are used for verification and tuning of the models. A simplified model is applied with random fields that obey a known kinetic energy spectrum. The periodization methods under consideration are detrending, the discrete cosine transform and the application of an extension zone. For the extension zone, three versions are applied: the Boyd method, the ALADIN method and the HIRLAM method. The results show that detrending and the discrete cosine transform have little impact on the spectra, as does the Boyd method for extension zone. For the ALADIN and HIRLAM methods, the impact depends on the width of the extension zone – the wider the zone, the more artificial energy and the larger impact on the spectra. The width of the extension zone correlates to the modifications in the shape of the spectra as well as to the amplitudes of the additional energy in the spectra.

Download
Short summary
The paper deals with the comparison of the common periodization methods used to obtain spectral fields of limited-area models for numerical weather prediction. The focus is on the impact the methods have on the spectra of the fields. The results show that detrending and DCT have little impact on the spectra, as does the Boyd method for extension zone. For the ALADIN and HIRLAM methods, the impact depends on the width of the extension zone - the wider the zone, the larger impact on the spectra.