基于GRAPES对流尺度模式一次低温事件的预报能力评估

Assessment of Prediction on a Low Temperature Event Based on the GRAPES Convection-Scale Model

摘要: 为评估GRAPES对流尺度模式资料对于低温事件的预报能力，本文选择2021年1月4日至1月12日我国黄山及周边区域，ERA5的2 m温度再分析资料、GRAPES对流尺度3 km的2 m温度资料，并使用图像识别的特征区域提取算法，将低于5℃的区域提取出来，作为本文的主要研究对象，通过分析GRAPES模式预报的低温及其分布、面积、区域平均温度，与ERA5的比较以评估其预报能力，结果表明：1) GRAPES对于所选个例预报较为准确，在个别时间有一定偏差，对于个例的位置和时间变化特征预报较好；2) GRAPES预报相对于ERA5资料，低温区域的位置分布比较一致，部分时间段预报偏南；3) 低温区域的面积，在个例出现时其相关系数较高，预报较为准确；4) 区域平均温度预报效果较好，且整体的变化趋势一致，区域平均温度的误差在2℃以内；5) 区域最低温度的位置GRAPES预报相较ERA5偏南，在寒潮过程中，预报的最低位出现的位置一致，GRAPES对于最低温的预报值整体偏低；6) GRAPES模式和ERA5的日变化特征比较显示，GRAPES预报的升降温时间比ERA5落后1小时。

Abstract: In order to evaluate the prediction ability of GRAPES convective scale model data for low temperature events, this paper selects the 2 m temperature reanalysis data of ERA5 and the GRAPES convective scale 3 km temperature data from January 4 to January 12, 2021 in Huangshan Mountain and its surrounding areas in my country. And we use the feature area extraction algorithm of image recognition to extract the area below 5˚C, as the main research object of this paper, analyze the low temperature predicted by GRAPES model and its distribution, area, and regional average temperature. Assessing its forecasting ability, the results show that: 1) GRAPES is more accurate for the selected individual cases, although there are certain deviations in individual times, but the location and time variation characteristics of individual cases are better predicted. 2) Compared with the ERA5 data, the GRAPES forecast is relatively consistent in the location distribution of the low temperature area, and some time periods are forecast to be southerly. 3) For the area of the low temperature area, the correlation coefficient is higher when a single case occurs, and the forecast is more accurate. 4) The regional average temperature forecast effect is good, and the overall change trend is consistent. From the appearance of the individual cases to the disappearance, the error of the regional average temperature is within 2˚C. 5) The location of the regional minimum temperature in the GRAPES forecast is more souther than that of ERA5, but during the cold wave process, the forecasted minimum location is relatively consistent, but the GRAPES forecast value for the minimum temperature is generally lower. 6) The comparison of the diurnal variation characteristics of the GRAPES model and ERA5 shows that the warming and cooling time predicted by GRAPES is 1 hour behind that of ERA5.