Abstracts for the 6th International GAME Conf.

3-5 December 2004

Kyoto Japan

Predictability of the Madden and Julian Oscillation in JMA one-month forecasts

Takuji Kubota (1), Hitoshi Mukougawa (1), Tatsuya Iwashima (1)

The Madden and Julian Oscillation (MJO) controls convection in the South Asia on time scales of 30-70 days, and it is typically associated with monsoon rainfall over northern Australia. In this work, the predictability of the MJO is investigated using velocity potential on 200 hPa surface of operational Japan Meteorological Agency (JMA) one-month forecasts in three recent Northern Hemisphere winter seasons during 2001-2004. The one-month forecasts are produced every Wednesday and Thursday based upon a JMA operational model (GSM0103) with T106 and 40 vertical levels. Anomaly correlation (AC) in the tropics (30S-30N), which is used as a measure of forecast skill, is shown to be above 0.6 and 0.3 until 5 and 11 days lead-time, respectively. The AC of the 2003-2004 season is highest in the three seasons. Root mean square errors for projection of forecasts on dominant modes in the tropics, which correspond to the MJO, are below the observed standard deviations until 10-13 days lead-time. Errors of amplitude and phase can be separately calculated through transformation to cylindrical coordinates using first two principal component time series and the projection time series of the forecast on the principal modes. Mean square errors normalized by the observed variances for the phase increase more rapidly and strongly than those for the amplitude. In earlier lead-times, the AC is positively correlated with the MJO amplitude although not very clear. No significant correlation is found after the middle stage of forecasts. A large difference between the AC values with the initial conditions on Wednesday and Thursday is found during 13-21 days lead-time for runs with initial dates of 10th/11th March 2004. This is related to the representation of the eastward propagating disturbance over the tropical eastern Pacific.

Submittal Information

Name : Date :
    Takuji Kubota
    29-Jul-04-17:01:22
Organization : Theme :
    Disaster Prevention Research Institute, Kyoto University
    Theme 4
Address : Presentation :
    611-0011
    Poster or oral
Country : Abstract ID :
    Japan
    T4TK29Jul04170122
Phone : Fax :
    +81-774-38-4155
    +81-774-38-4153
E-mail :
    kubota@dpac.dpri.kyoto-u.ac.jp