This paper considers parameter choice problems in numerical weather prediction system. A global optimization method is proposed to choose model parameters according to observation datas. At first, the parameter choice problem is transformaed into an inverse problem of a partial differential equation. and a suitable objetive function can be established accordingly. Then the global minimizer of the objective function is searched to get satisfied parameter estimate values. Numerical experiments have verified the effectiveness and wide suitability of the method.
展开▼