Analysis of the surface temperature and wind forecast errors of the NCAR-AirDat operational CONUS 4-km WRF forecasting system

Investigating the characteristics of model-forecast errors using various statistical and object-oriented methods is necessary for providing useful guidance to end-users and model developers as well. To this end, the random and systematic errors (i.e., biases) of the 2-m temperature and 10-m wind predictions of the NCAR-AirDat weather research and forecasting (WRF)-based real-time four-dimensional data assimilation (RTFDDA) and forecasting system are analyzed. This system has been running operationally over a contiguous United States (CONUS) domain at a 4-km grid spacing with four forecast cycles daily from June 2009 to September 2010. In the result an exceptionally useful forecast dataset was generated and used for studying the error properties of the model forecasts, in terms of both a longer time period and a broader coverage of geographic regions than previously studied. Spatiotemporal characteristics of the errors are investigated based on the 24-h forecasts between June 2009 and April 2010, and the 72-h forecasts between May and September 2010. It was found that the biases of both wind and temperature forecasts vary greatly seasonally and diurnally, with dependency on the forecast length, station elevation, geographical location, and meteorological conditions. The temperature showed systematic cold biases during the daytime at all station elevations and warm biases during the nighttime above 1,000 m above sea level (ASL), while below 600 m ASL cold biases occurred during the nighttime. The forecasts of surface wind speed exhibited strong positive biases during the nighttime, while the negative biases were observed in the spring and summer afternoons. The surface wind speed was mostly over-predicted except for the stations located between 1,000 and 2,100 m ASL, for which negative biases were identified for most forecast cycles. The highest wind-speed errors were found over the high terrain and near sea-level stations. The wind-direction errors were relatively large at the high-terrain elevation in the Rocky and Appalachian mountain ranges and the western coastal areas and the error structure exhibited notable diurnal variability.     

Verification of the Weather Research and Forecasting Model when Forecasting Daily Surface Conditions in Southern Alberta

The Weather Research and Forecasting (WRF) model was compared with daily surface observations to verify the accuracy of the WRF model in forecasting surface temperature, pressure, precipitation, wind speed, and direction. Daily forecasts for the following two days were produced at nine locations across southern Alberta, Canada. Model output was verified using station observations to determine the differences in forecast accuracy for each season.

Although there were seasonal differences in the WRF model, the summer season forecasts generally had the greatest accuracy, determined by the lowest root mean square errors, whereas the winter season forecasts were the least accurate. The WRF model generally produced skillful forecasts throughout the year although with a smaller diurnal temperature range than observed. The WRF model forecast the prevailing wind direction more accurately than other directions, but it tended to slightly overestimate precipitation amounts. A sensitivity analysis consisting of three microphysics schemes showed relatively minor differences between simulated precipitation as well as 2?m surface temperatures.     

基于数值模式的月尺度近地层气象要素预报技术研究

利用WRF模式对美国NCEP发布的CFS气候预测业务产品在中国区域内进行动力降尺度预报,可得到预报时效为45天的逐6小时、30 km分辨率基础气象要素预测产品。再利用全国气象站观测资料和3个风电场70 m高度风速、温度观测资料对2015年冬季预测结果进行检验评估和分析,最后通过线性方法对地面要素预测结果和70 m高度风速、温度预测结果进行统计订正。结果表明:（1）2 m温度和相对湿度的全国预报平均绝对误差分别为4.71 ℃和18.81%,在华东、华中和华南地区误差较小;（2）10 m风速预报平均绝对误差为2.42 m/s,在东北、华北和西北地区误差较小;（3）线性订正后,2 m气温、相对湿度和10 m风速的预报绝对误差分别减小1.05 ℃、5.29%和1.47 m/s,并且订正后误差随时间变化更平稳;（4）订正后70 m高度风速和温度的预报绝对误差均减小,风速平均误差减小最大可达1.29 m/s（B塔）,气温平均绝对误差减小最大可达3 ℃（C塔）。研究结果表明,基于CFS产品和WRF模式的、与月尺度风电预报关系密切的气象要素预报性能较好,未来可将该方法尝试于风电场的月尺度功率预测产品研发。      

Evaluation of atmospheric boundary layer characteristics simulated by the regional model

Carried out is the comparison of the temporal courses of temperature and wind speed at different levels as well as of the wind and temperature profiles in the atmospheric boundary layer obtained from the WRF regional model forecasts and using the upper-air in situ and remote measurements in Moscow region. The errors in temperature and wind speed forecasts at different levels are computed as well as the statistical estimates of the forecast of temperature inversions, atmospheric stratification types, and monthly mean wind speed profiles on the basis of model forecasts and acoustic sounding.     

Intercomparison of the performance of MM5/WRF with and without satellite data assimilation in short-range forecast applications over the Indian region

The present study is conducted to verify the short-range forecasts from mesoscale model version5 (MM5)/weather research and forecasting (WRF) model over the Indian region and to examine the impact of assimilation of quick scatterometer (QSCAT) near surface winds, spectral sensor microwave imager (SSM/I) wind speed and total precipitable water (TPW) on the forecasts by these models using their three-dimensional variational (3D-Var) data assimilation scheme for a 1-month period during July 2006. The control (without satellite data assimilation) as well as 3D-Var sensitivity experiments (with assimilating satellite data) using MM5/WRF were made for 48 h starting daily at 0000 UTC July 2006. The control run is analyzed for the intercomparison of MM5/WRF short-range forecasts and is also used as a baseline for assessing the MM5/WRF 3D-Var satellite data sensitivity experiments. As compared to the observation, the MM5 (WRF) control simulations strengthened (weakened) the cross equatorial flow over southern Arabian sea near peninsular India. The forecasts from MM5 and WRF showed a warm and moist bias at lower and upper levels with a cold bias at the middle level, which shows that the convective schemes of these models may be too active during the simulation. The forecast errors in predicted wind, temperature and humidity at different levels are lesser in WRF as compared to MM5, except the temperature prediction at lower level. The rainfall pattern and prediction skill from day 1 and day 2 forecasts by WRF is superior to MM5. The spatial distribution of forecast impact for wind, temperature, and humidity from 1-month assimilation experiments during July 2006 demonstrated that on average, for 24 and 48-h forecasts, the satellite data improved the MM5/WRF initial condition, so that model errors in predicted meteorological fields got reduced. Among the experiments, MM5/WRF wind speed prediction is most benefited from QSCAT surface wind and SSM/I TPW assimilation while temperature and humidity prediction is mostly improved due to latter. The largest improvement in MM5/WRF rainfall prediction is due to the assimilation of SSM/I TPW. The assimilation of SSM/I wind speed alone in MM5/WRF degraded the humidity and rainfall prediction. In summary the assimilation of satellite data showed similar impact on MM5/WRF prediction; largest improvement due to SSM/I TPW and degradation due to SSM/I wind speed.     

Multiseason evaluation of the MM5, COAMPS and WRF over southeast United States

Three models, MM5, COAMPS, and WRF, have been applied for the warm season in 2003 and the cool season in 2003?C2004 to evaluate their performances. All models run over the same domain area covering the north Gulf Mexico and southeastern United States (US) region with the same spatial resolution of 27?km. It was found that the temporal variations of the mean error distribution and strength at 24 and 36?h were rather weak for surface temperature, sea level pressure, and surface wind speed for all models. A warm bias in surface temperature forecasts dominated over land during the warm season, whereas a cool bias existed during the cool season. The MM5 and WRF produced negative biases of sea level pressure during the warm season and positive biases during the cool season while the COAMPS yielded a similar distribution of sea level pressure biases during both seasons. During both seasons, similar surface wind speed biases produced by each model included a high wind speed forecast over most areas by MM5 while the COAMPS and WRF yielded weak surface winds over the western Plains and stronger surface winds over the eastern Plains. Root-mean-squared errors revealed that the forecast of surface temperature, sea level pressure, and surface wind speed were degraded with the increase of forecast time. For rainfall evaluation, it was found that the MM5 underpredicted seasonal precipitation while the COAMPS and WRF overpredicted. The bias scores revealed that the MM5 yielded an underprediction of the coverage of precipitation areas, especially for heavier rainfall events. The MM5 presented the lower threat score at lighter rainfall events compared to the COAMPS and WRF. For moderate and heavier thresholds, all models lacked forecast accuracy. The WRF accuracy in predicting precipitation was heavily dependent upon the performance of the selected cumulus parameterization scheme. Use of the Grell?CDevenyi and Bette?CMiller?CJanjic schemes helps suppress precipitation overprediction.     

An Experiment on the Prediction of the Surface Wind Speed in Chongli Based on the WRF Model: Evaluation and Calibration

In this study,the ability of the Weather Research and Forecasting(WRF)model to generate accurate near-surface wind speed forecasts at kilometer-to subkilometer-scale resolution along race tracks(RTs)in Chongli during the wintertime is evaluated.The performance of two postprocessing methods,including the decaying-averaging(DA)and analogy-based(AN)methods,is tested to calibrate the near-surface wind speed forecasts.It is found that great uncertainties exist in the model’s raw forecasts of the near-surface wind speed in Chongli.Improvement of the forecast accuracy due to refinement of the horizontal resolution from kilometer to subkilometer scale is limited and not systematic.The RT sites tend to have large bias and centered root mean square error(CRMSE)values and also exhibit notable underestimation of high-wind speeds,notable overestimation or underestimation of the near-surface wind speed at high altitudes,and notable underestimation during daytime.These problems are not resolved by increasing the horizontal resolution and are even exacerbated,which leads to great challenges in the accurate forecasting of the near-surface wind speed in the competition areas in Chongli.The application of postprocessing methods can greatly improve the forecast accuracy of near-surface wind speed.Both methods used in this study have comparable abilities in reducing the(positive or negative)bias,while the AN method is also capable of decreasing the random error reflected by CRMSE.In particular,the large biases for high-wind speeds,wind speeds at high-altitude stations,and wind speeds during the daytime at RT stations can be evidently reduced.     

2013年华中区域中尺度业务数值预报的客观检验

应用国家气象中心模式检验方法,对华中区域中尺度业务数值预报模式WRF和中国气象局下发的GRAPES模式2013年1-12月的预报产品（包括降水、2 m温度场、850 hPa温度场、850 hPa风场和500 hPa位势高度场预报）进行统计检验。检验结果表明：所有要素24 h预报均优于48 h;对于晴雨预报,GRAPES模式TS评分高于WRF,但随着降水量级增大,WRF的TS评分基本都高于GRAPES,同时WRF降水预报范围明显偏大;分析2 m温度场的均方根误差及预报准确率发现,WRF的2 m温度场预报优于GRAPES,且暖季预报优于冷季;形势场要素分析表明,WRF对850 hPa温度场和风场预报具有相当大的优势,全年误差变化较稳定,而对500 hPa位势高度场的预报误差存在一定的季节性特征,即夏半年WRF高度场预报优于冬半年,GRAPES模式则相反。总体上,华中区域中尺度业务数值模式产品对天气预报具有一定的参考价值。     

Impacts of Land-Use Data on the Simulation of Surface Air Temperature in Northwest China

This study examines the impacts of land-use data on the simulation of surface air temperature in Northwest China by the Weather Research and Forecasting (WRF) model. International Geosphere–Biosphere Program (IGBP) landuse data with 500-m spatial resolution are generated from Moderate Resolution Imaging Spectroradiometer (MODIS) satellite products. These data are used to replace the default U.S. Geological Survey (USGS) land-use data in the WRF model. Based on the data recorded by national basic meteorological observing stations in Northwest China, results are compared and evaluated. It is found that replacing the default USGS land-use data in the WRF model with the IGBP data improves the ability of the model to simulate surface air temperature in Northwest China in July and December 2015. Errors in the simulated daytime surface air temperature are reduced, while the results vary between seasons. There is some variation in the degree and range of impacts of land-use data on surface air temperature among seasons. Using the IGBP data, the simulated daytime surface air temperature in July 2015 improves at a relatively small number of stations, but to a relatively large degree; whereas the simulation of daytime surface air temperature in December 2015 improves at almost all stations, but only to a relatively small degree (within 1°C). Mitigation of daytime surface air temperature overestimation in July 2015 is influenced mainly by the change in ground heat flux. The modification of underestimated temperature comes mainly from the improvement of simulated net radiation in December 2015.     

一次粤西南暴雨过程的预报误差来源分析

本文基于WRF模式研究了2015年5月16～17日广东西南地区的一次暴雨过程的预报误差来源。首先比较了以NCEP_FNL为初始资料的WRF模式的模拟预报（记为WRF_FNL）和ECMWF（European Centre for Medium-Range Weather Forecasts）关于该次暴雨过程的确定性预报。结果表明,ECMWF具有较高的预报技巧,因此,认为ECMWF的模式和初始场都较为准确。进一步,以ECMWF的初值作为初始场,选用相同的物理参数化方案,再次用WRF模式进行预报（预报结果记为WRF_EC）。结果表明相对WRF_FNL,WRF_EC的预报结果有明显改善。这表明,初始场的改进对预报有较大的影响,初始误差是预报误差的重要来源。进一步,分析了初始误差的主要来源区域和来源变量。结果表明,南海北部湾至广西西南区域为本次暴雨预报初始误差的主要来源区域,而初始温度场和初始湿度场则为此次暴雨预报初始误差的主要来源变量。同时改进初始温度场和湿度场可以较大程度提高本次暴雨过程的预报技巧。     

ERA5再分析资料对中国大陆区域近地层风速气候特征及变化趋势再现能力的评估

基于台站观测资料,评估了欧洲中期天气预报中心(ECMWF)最高时空分辨率的第五代大气再分析资料(ERA5)对1979～2018年间中国大陆区域10 m高度风速的气候特征及其变化趋势的再现能力,并同步对比分析了ERA5资料100 m高度风速的特征和长期趋势.结果表明,ERA5资料10 m和100 m风速在空间分布、年—季...     

VALIDATION OF NEAR-SURFACE WINDS OBTAINED BY A HYBRID WRF/CALMET MODELING SYSTEM OVER A COASTAL ISLAND WITH COMPLEX TERRAIN

The results from a hybrid approach that combines a mesoscale meteorological model with a diagnostic model to produce high-resolution wind fields in complex coastal topography are evaluated.The diagnostic wind model(California Meteorological Model,CALMET) with 100-m horizontal spacing was driven with outputs from the Weather Research and Forecasting(WRF) model to obtain near-surface winds for the 1-year period from 12 September 2003 to 11 September 2004.Results were compared with wind observations at four sites.Traditional statistical scores,including correlation coefficients,standard deviations(SDs) and mean absolute errors(MAEs),indicate that the wind estimates from the WRF/CALMET modeling system are produced reasonably well.The correlation coefficients are relatively large,ranging from 0.5 to 0.7 for the zonal wind component and from 0.75 to 0.85 for the meridional wind component.MAEs for wind speed range from 1.5 to 2.0 m s-1 at 10 meters above ground level(AGL) and from 2.0 to 2.5 m s-1 at 60 m AGL.MAEs for wind direction range from 30 to 40 degrees at both levels.A spectral decomposition of the time series of wind speed shows positive impacts of CALMET in improving the mesoscale winds.Moreover,combining the CALMET model with WRF significantly improves the spatial variability of the simulated wind fields.It can be concluded that the WRF/CALMET modeling system is capable of providing a detailed near-surface wind field,but the physics in the diagnostic CALMET model needs to be further improved.     

A Deep Learning Method for Bias Correction of ECMWF 24–240 h Forecasts

Correcting the forecast bias of numerical weather prediction models is important for severe weather warnings. The refined grid forecast requires direct correction on gridded forecast products, as opposed to correcting forecast data only at individual weather stations. In this study, a deep learning method called CU-net is proposed to correct the gridded forecasts of four weather variables from the European Centre for Medium-Range Weather Forecast Integrated Forecasting System global model(ECMWF-IFS): 2-m temperature, 2-m relative humidity, 10-m wind speed, and 10-m wind direction, with a forecast lead time of 24 h to 240 h in North China. First, the forecast correction problem is transformed into an image-toimage translation problem in deep learning under the CU-net architecture, which is based on convolutional neural networks.Second, the ECMWF-IFS forecasts and ECMWF reanalysis data(ERA5) from 2005 to 2018 are used as training,validation, and testing datasets. The predictors and labels(ground truth) of the model are created using the ECMWF-IFS and ERA5, respectively. Finally, the correction performance of CU-net is compared with a conventional method, anomaly numerical correction with observations(ANO). Results show that forecasts from CU-net have lower root mean square error, bias, mean absolute error, and higher correlation coefficient than those from ANO for all forecast lead times from 24 h to 240 h. CU-net improves upon the ECMWF-IFS forecast for all four weather variables in terms of the above evaluation metrics, whereas ANO improves upon ECMWF-IFS performance only for 2-m temperature and relative humidity. For the correction of the 10-m wind direction forecast, which is often difficult to achieve, CU-net also improves the correction performance.     

概率密度匹配方法在我国近海海面10m风速预报中的应用

海面风速对航运及海上生产作业影响重大,但数值模式对于海面的风速预报仍存在较大误差.为降低数值模式海面10 m风速预报的系统性误差,提高海上大风预报准确率,基于2017-2019年中国气象局地面气象观测资料对ECMWF确定性模式的10 m风场预报结果进行检验评估,并采用概率密度匹配方法对模式误差进行订正.分析结果表明,概...     

华南区域高分辨率数值模式前汛期预报初步评估

基于华南地区自动站逐小时观测资料, 采用传统站点评分、邻域法等评估华南区域高分辨率数值模式(包括GRAPES_GZ_R 1 km模式和GRAPES_GZ 3 km模式)对降水、地面温度和风场等要素的预报能力。结果表明: GRAPES_GZ_R 1 km模式的降水预报技巧优于GRAPES_GZ 3 km模式, 模式预报以正偏差为主。对于不同起报时间的预报, 00时(世界时, 下同)起报的预报效果优于12时。GRAPES_GZ_R 1 km模式的TS评分是GRAPES_GZ 3 km模式的两倍以上, 对不同降水阈值的评分均较高。分数技巧评分(FSS)显示GRAPES_GZ_R 1 km模式6 h累计降水预报在0.1 mm、1 mm及5 mm以上的降水均可达到最低预报技巧尺度, 对所检验降水对象的空间位置把握能力更好。2 m气温和10 m风速检验结果表明两个模式均能较好把握广东省温度的分布特征, GRAPES_GZ_R 1 km模式对2 m气温预报结果优于GRAPES_GZ 3 km模式, 预报绝对误差更小; 两个模式对风速的预报整体偏强, 预报偏差在1~4 m/s之间, 但相比之下GRAPES_GZ 3 km模式在风场预报上表现更好。GRAPES_GZ_R 1 km模式的2 m气温和10 m风速预报偏差随降水过程存在明显波动, 强降水过后温度预报整体偏低, 风速预报偏强, 在模式产品订正、使用等需要考虑模式对主要天气系统的预报情况。总的来说, GRAPES_GZ_R 1 km模式的预报产品具有较好的参考价值。      

BJ-RUC系统模式地面气象要素预报效果评估

利用自动气象站逐小时地面观测资料,采用客观检验方法对北京市气象局快速更新循环预报 (BJ-RUC) 系统在2008—2010年5—9月的预报结果进行检验,初步评估了BJ-RUC系统对地面气象要素的业务预报性能。结果表明:BJ-RUC系统对地面气象要素预报与实况的变化趋势有很好的一致性。其中,2 m温度预报整体偏高,误差范围为-1.5~1.5℃,早上和傍晚偏大,正午偏小;2 m相对湿度的预报整体偏低,误差为-25%~0,白天偏大,夜间偏小;10 m风速预报明显偏大,午后尤为显著,误差为0.6~1.2 m·s-1;6 h累积降水的晴雨预报效果较好,TS评分可达到0.4。系统在初始起报时次的稳定性较差,从第3个起报时次开始逐渐稳定,但预报误差随着预报时效的增长逐渐增大,12 h内的预报误差较小,预报结果较可靠,在短时临近预报中具有参考价值。     

风电场风机覆冰期预报方法

采用Makkonen结冰增长模型镶嵌冰表面辐射融化模型建立适用于风机覆冰的预报模型，基于WRF耦合CALMET综合模式预测结果和风机性能参数驱动模型，进行风机覆冰开始和结束时间的预报。以湖北通山九宫山风电场为研究对象，对2012年12月至2013年12月期间监测到19次风机覆冰进行分析，分别采用提前1 d、提前2 d、提前3 d三种预报方案开展覆冰起止时间的预报，发现该模型对风机覆冰具有一定的提前预警能力，且越临近覆冰发生时间，预报准确性越高。     

复杂下垫面风电场风速数值模拟及误差特征

利用WRF模式分别对沿海及山地条件下风电场风速进行高分辨数值模拟,并对其误差特征进行分析,结果表明:1)WRF模式对复杂地形条件下的风速模拟性能良好,模拟值较好地体现天气尺度的周期变化;2)沿海及山地条件下模拟与观测的误差特征各不相同。模式静态数据未能显现沿海的小岛,并且低估了山地测风塔所在的海拔,导致沿海平均模拟风速偏大,山地平均模拟风速偏小;3)分析不同风向的归一化均方根误差,沿海陆风情况下,下垫面相对复杂,误差明显增大;沿海海风情况下,下垫面均一,误差明显减小;4)仅作单个风电场周边数百平方千米的模拟,采用一台12核的服务器进行WRF模式的并行计算可满足48 h短期预测的时效性。仅仅提高模拟的网格分辨率,并不一定能提升模拟的准确性。     

单多普勒天气雷达反演风场高原低纬区域同化应用

为了改善低纬高原地区天气预报水平,利用WRF(Weather Research and Forecasting)模式及其变分同化系统进行雷达VAD(Velocity Azimuth Display)反演风场资料同化试验。通过设计不同的试验方案,对2009年6月30日00:00至7月1日00:00发生在云南的一次强降水过程进行数值模拟和对比分析,结果表明:同化VAD反演风场资料后对区域模式的风矢量初始场有明显影响。同化系统能把雷达反演风场信息有效地引入模式初始场,改善强降水区域的水汽输送和风场辐合强度;同化VAD反演风场资料后对区域模式累计降水预报有一定改进作用。从长时间累计降水量定量检验结果看,具体表现为25mm以上量级的降水准确率明显提高、漏报率下降,预报偏差更趋合理。不同的同化试验方案之间的模拟结果差异较大。同化频率越高、同化持续时间越长,对区域模式初始场和预报场的影响越明显。但同化持续时间不宜过长,否则可能导致系统移速过快、降水强度偏大、空报率增加等异常。     

大小兴安岭林区风场动力降尺度对比研究

对2017年春季黑龙江省大、小兴安岭林区的6个代表站点10 m风场进行降尺度分析,并结合观测数据对比分析了WRF模式和CALMET降尺度模式的10 m风速、风向预报结果。结果表明：两模式逐小时风速预报与观测的相关系数为0.5-0.7,且随着风速的增加,模式的预报准确率逐渐提高,夜间的风速预报偏差较大,进入白天后,偏差明显减小。WRF模式对风速变化趋势的预报效果优于CALMET模式,与观测的风速相关性更高,而CALMET模式对较大风速的预报效果优于WRF模式。在风向预报方面,WRF和CALMET的风向模拟与观测风向均有较好的一致性,模式预报准确率较高的两个风向也刚好对应各站的盛行风向。同时,本文用回归方法对日平均风速进行订正发现,订正后各站的日平均风速预报准确率平均提高了50%,具有较好的业务应用价值。