4.2 Basic Characteristics of Stratospheric Predictability in One-month Ensemble Hindcast Experiments for 1979-2009

Tuesday, 18 June 2013: 3:30 PM
Viking Salons DE (The Hotel Viking)
Masakazu Taguchi, Aichi University of Education, Kariya, Japan

This study investigates basic characteristics of predictability (PR) especially for NH winter stratosphere by comparing the JRA-25/JCDAS reanalysis data to 1-month ensemble hindcast (HC) experiment data. The HC experiments were conducted by the JMA using March, 2011 version of the 1-month ensemble prediction system. The ensemble predictions were made from each of the 10th, 20th, and last day of each month for 1979-2009, with an ensemble size of five.

First, we characterize the HC data from several basic aspects such as systematic biases, spread, and forecast skill (root mean square of geopotential height error). This characterization includes contrasting the stratosphere and troposphere, as well as the different seasons. The PR of the NH winter stratosphere is characterized by a longer predictable period and its larger variability than that of the troposphere.

Second, we further explore in detail the PR of the NH winter stratosphere in two respects. We examine initial conditions for which the HC data exhibits large spread and low skill: it turns out that such conditions are when the polar night jet is in the phase of weakening. We also examine the relationship between the spread and skill: the two quantities are correlated as expected (large spread corresponds to low skill), but also their distributions are characterized by outliers which have markedly low skill for given spread.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner