基于江西省站点资料对智能网格实况分析产品的检验评估

The Quality Assessment of the Intelligent Grid Real-Time Analysis Products Using the Station Data in Jiangxi Province

利用江西省2019年3月~2020年2月气象站观测资料,在小时尺度上,对中国气象局下发的智能网格实况的2 m气温、湿度、风速、降水等多个气象要素产品进行检验评估.研究分析各类产品的数据误差及其时空变化规律,验证该产品在研究区的适用性.研究结果表明:2 m气温、相对湿度产品与站点观测较为一致,其相关系数均可达0.9以上,2 m气温均方根误差在1.0℃左右,准确率可达80%以上,相对湿度均方根误差小于10%,准确率可达90%以上.风速产品的数据精度较低,相比站点观测风速明显偏弱,70%站点的风向准确率小于50%,风向准确率仅为40%左右.降水产品能准确反映年内逐小时变化趋势,但存在一定程度低估,相关系数基本在0.8以上,平均误差为-0.1~0.1 mm·h-1,平均绝对误差小于0.2 mm·h-1.与站点观测相比,降水产品对强降水明显偏弱.地形对实况网格产品准确性有较大影响,产品误差在地形复杂山区中比在平原地区中明显偏大.总体上,2 m气温、相对湿度产品质量较高,基本可满足业务服务需求.风速产品及降水产品在大降水量级下的误差较大,与业务服务需求有一定差距.

In the paper,the quality of intelligent grid real-time analysis products that consist of the 2 m temperature,humidity,wind speed,precipitation issued by China Meteorological Administration is assessed by the meteorological station observation data from January 2019 to December 2019 in Jiangxi province.The data error of each products and its temporal and spatial variation characteristic were analyzed to verify the applicability of products in this study area.Results show that the 2 m temperature and relative humidity products have well consistency with the station observation data.The correlation coefficients are all above 0.9,the RMSE of the 2 m temperature is about 1.0℃,and the accuracy is above 80%.The RMSE of the relative humidity is less than 10%,and the accuracy is above 90%.The data accuracy of the wind speed product is a bit low.The wind speed is obviously weak compared with the site observation wind speed.70%of the station wind direction accuracy is less than 50%,while the wind direction accuracy is only about 40%.Precipitation products can accurately reflect the trend of hourly change in the year,but there is a certain underestimation.The correlation coefficient is mainly above 0.8,the average error is between-0.1 mm·h-1 and 0.1 mm·h-1,and the average absolute error is less than 0.1 mm·h-1.Compared with station observation,precipitation products are obviously weak at heavy rainfall.The terrain has a great influence on the accuracy of the real-time grid products.The product error is obviously larger in the complex terrain mountain area than in the plain area.On the whole,the quality of 2 m temperature and relative humidity products is high and basically can meet the application and service needs.Wind speed products and precipitation products at the large precipitation level have large errors,and there is a certain gap for the application and service needs.