精细化逐时滚动温度预报方法及检验

以Grapes数值模式预报为基础,首先利用卡尔曼滤波方法对Grapes模式的温度预报进行释用,再将模式统计输出方法应用于卡尔曼滤波结果,从而得到站点逐时滚动温度预报,最后通过站点-格点映射方法将站点预报误差反馈到最匹配的格点上,实现精细化逐时滚动温度预报(SHUF)。检验结果表明,Grapes模式的24小时温度预报CSI评分稳定在0.4左右;卡尔曼滤波方法的CSI评分介于0.47～0.43之间;而SHUF的CSI评分在1～6小时内由0.91降至0.64,7～16小时的CSI评分由0.6逐渐降低至0.52,17～24小时的CSI介于0.5～0.45之间,均优于同期Grapes模式预报和卡尔曼滤波释用结果。精细化逐时滚动温度预报方法利用最新的气象观测要素对数值模式预报的结果进行订正,可有效改进数值模式的短时温度预报能力。

RESEARCH OF SEAMLESS HOURLY UPDATED FORECAST OF TEMPERATURE AND VERIFICATION

A short-term temperature forecast guidance,called seamless hourly updated forecast(SHUF) for temperature,is developed in this paper.First,based on the result of Global and Regional Assimilation and Prediction System(GRAPES) model,the Kalman filter is used to generate preparatory interpretational forecast guidance.Second,the model output statistics(MOS) method is applied to generate an hourly-updated forecast guidance.Finally,the SHUF is provided by feeding back the site prediction error to the best match grid points.The verification has shown improvements in the SHUF:the GRAPES model shows a steady CSI score of 0.4 in 1～24 h projections,while the guidance with the Kalman filter shows descending CSI scores from 0.47 to 0.43.The SHUF guidance shows CSI scores decreasing from 0.91 to 0.64 in 1～6 h projections,and the score gradually decreases from 0.6 to 0.52 in 7～16 h projections.Improvement is also shown with scores of 0.5 to 0.45 in 17～24 h projections.Using the latest meteorological factors,the SHUF guidance can effectively improve the skill in short-term temperature forecasting.