中国近地面气温直减率的时空分布及其在NCEP气温预报误差订正中的应用

Spatiotemporal distribution of near-surface air temperature lapse rate in China and its application in bias correction of NCEP temperature forecasts

摘要:

近地面气温直减率受地形地貌、地理位置、季节变化、昼夜变化及人类活动等因素的显著影响,在中国区域使用单一的气温直减率不能准确表征其时空特征。本文基于中国2 427个国家基本气象站2013年3月1日—2022年2月28日的逐小时气温观测数据,利用三元线性回归方程,分区域、季节、昼夜拟合出近地面(2 m)气温直减率,并将其应用于NCEP气温预报产品的高度订正中以验证其可靠性。结果表明:1)从全国区域平均来看,近地面气温直减率年平均为0.57 ℃/(100 m),白天略高于夜晚。2)近地面气温直减率季节及昼夜差异较大,夏季最高,平均为0.63 ℃/(100 m);冬季最低,平均为0.47 ℃/(100 m)。昼夜最大差异在春夏秋冬4个季节分别为0.22、0.29、0.28、0.50 ℃/(100 m)。3)应用统计出的气温直减率对NCEP气温预报进行高度订正,发现全国平均的气温平均绝对误差较订正前在未来360 h各个预报时效均有降低,最大降幅可达1.20 ℃。对于大多数子区域,夏季订正效果最好,白天比夜间订正效果要好。

Abstract:

The near-surface air temperature lapse rate (NTLR) is a crucial yet complex meteorological parameter that determines the rate at which air temperature decreases with increasing altitude.A uniform lapse rate across the troposphere cannot adequately capture the spatiotemporal variability of NTLR.Given the significant influences of topography,geographical location,seasonal cycles,diurnal variations,and human activities on NTLR,this study highlights the need for a spatiotemporally specific approach to accurately characterize these variations.The study has two primary objectives:first,to analyze the spatial and temporal variability of NTLR across different regions of China,and second,to assess the impact of these variations on the correction of temperature forecast biases in the NCEP model.According to comprehensive physical regionalization,China is divided into 33 sub-regions.Utilizing a dataset of hourly temperature observations from 2 427 national-level meteorological stations in China,covering the period from March 1,2013,to February 28,2022,the study calculates regional and seasonal NTLR values for both day and night using a ternary linear regression model.This model accounts for altitude,latitude,and longitude,providing a detailed representation of the NTLR’s relationship with geographical factors.The calculated NTLR values are then applied to correct elevation-based biases in the NCEP temperature forecasts from March 1,2022,to February 28,2023.The results indicate that:1) Significant spatiotemporal variations in NTLR exist across China,with differences occurring even within the same sub-region.The annual mean NTLR for China is 0.57 ℃/(100 m),with nighttime values slightly higher than daytime values.2) Seasonal and diurnal variations in NTLR are substantial,with the highest rates observed in summer (average 0.63 ℃/(100 m)) and the lowest in winter (average 0.47 ℃/(100 m)).The maximum day-night differences in NTLR are 0.22,0.29,0.28,and 0.50 ℃/(100 m) for spring,summer,autumn,and winter,respectively.3) Applying the statistically derived NTLR to correct elevation-based biases in the NCEP temperature forecasts reduces the mean absolute error (MAE) across China for all forecast periods up to 360 h,with an average reduction of 0.56 ℃ and a maximum reduction of 1.20 ℃.The correction is most effective in summer and during the daytime across most sub-regions.However,in some areas,such as the central and northern Qinghai-Xizang Plateau,applying NTLR for bias correction has led to reduced forecast accuracy,indicating that NTLR application must be adapted to local conditions.Future research should explore the impact of varying altitude ranges,slope aspects,and other factors on NTLR within the context of comprehensive physical regionalization.