JAMSTEC > HPCI SPIRE Field3 > R&D Project > Ultra-high Precision Meso-Scale Weather Prediction


R&D Project

  Ultra-high Precision Meso-Scale Weather Prediction

     Managing Director:
     Kazuo Saito
Ensemble forecast experiments on tornadoes that occurred in Tsukuba on May 6, 2012

 Accuracy of numerical weather prediction (NWP) has been remarkably improved in recent years, but precise prediction of severe meteorological phenomena such as torrential rains and local heavy rainfalls is still a difficult and challenging subject due to the following reasons.

1) Accuracy of initial condition is insufficient for small spatial scale of the severe phenomena.
2) Mesoscale convective systems are often very sensitive to small perturbations of the initial condition and computational conditions.
3) Cumulonimbus is not fully resolvable in the horizontal resolution of the current numerical weather prediction systems.

Data assimilation and the ensemble forecast with the cloud-resolving resolution are required to overcome above problems, and the computational resource is a key to reduce the compromise of the resolutions and the number of ensemble members. In this subject, we will perform following three subjects using the “K Computer” and show feasibility of precise prediction of severe mesoscale phenomena by a cloud-resolving NWP system.

  Research Goal

  Results of Researches