Application of Condition Number to Regional Mean Rainfall Forecast
Aim to the frequent heavy rain and flood in North of Guangxi, using numerical weather forecast products of T213 model and Japan fine-mesh rainfall forecast model, a regional mean rainfall regressive forecast equation for north of Guangxi is set up on the basis of the calculation result of condition number. By an operational test in May and June 2004—2005, the errors of new forecast equation and traditional stepwise regression equation are 6.3569mm, 7.0096mm respectively, and the error of T213 in the corresponding period is 7.9456mm. The forecast capability of the new equation is better than the traditional regressive equation and T213 model, and it can be used widely in the operational weather forecast. Comparing the condition number of two forecast equations, the result shows that the collinearity of the new forecast equation is lower than the traditional stepwise regression equation, whose serious collinearity influences the forecast capability.