2014, 42(3):443-450.
Abstract:
Based on the observed meteorological data of Enshi Station from 2008 to 2011, numerical forecast products, and superior station guidance, by means of the combined method of M (numerical model forecast), E (learning experience of weather), D (diagnostic analysis), the predictors with atmospheric physics significance are designed from the aspect of the influencing factors of temperature changes, such as the atmospheric stability, temperature advection, water vapor conditions, etc.; or a combination of factors according to the needs is considered, and the superior guidance products are used as predictors directly. Using the conventional statistical forecast method (stepwise regression), taking high and low temperature as predictands, the local temperature prediction model is established considering the direct factors influencing the atmospheric temperature. The local temperature MOS forecast model is established after sky condition classification, and the numerical forecast products are selected corresponding to the appearing time of high and low temperature, which is of significance to the quality improvement of local temperature forecasting. During the model building, the combined method and reprocessed numerical forecast factors are used, and the influence of weather variation on air temperature is considered, so the application and interpretation capability of numerical forecast products is enhanced. The comprehensive MOS forecasting referring to the objective numerical model products and superior guidance products is an attempt to improve the local weather forecast accuracy effectively. Test results also show that the local temperature MOS forecast method performed well, with the effectiveness significantly better than guidance forecasting.