Stochastic tendency in a global atmospheric model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014: 3:45 PM
Room C202 (The Georgia World Congress Center )
Myung-Seo Koo, Yonsei University, Seoul, South Korea; and S. Y. Hong

Handout (3.2 MB)

A new stochastic approach is suggested in order to improve weather prediction. Dynamical and physical model tendencies are perturbed by a random number that is sampled from a uniform distribution and a random interval that is dependent on the forecast time and vertical layer. Compared to a traditional approach of perturbing the physical tendency only, perturbing the dynamical tendency provides for greater improvement in forecast skill in terms of the 500-hPa geopotential height in a medium-range forecast. In addition, the best results are achieved by perturbing both the dynamical and physical tendencies simultaneously. This result implies that model uncertainties should be addressed in terms of not only physical parameterization but also the dynamical portion that used to be regarded as deterministically solved.

Supplementary URL: http://grims-model.org