祁连山综合观测网：兰州大学寒旱区科学观测网络CARN（瓜州站涡动相关仪-2018）

该数据集包含了2018年9月24日至2018年12月31日的兰州大学寒旱区科学观测网络瓜州站涡动相关仪观测数据。站点位于甘肃酒泉瓜州县柳园镇，下垫面是荒漠。观测点的经纬度是95.673E，41.405N，海拔2014m。涡动相关仪的架高4m，采样频率是10Hz，超声朝向是正北向，超声风速温度仪（CSAT3）与CO2/H2O分析仪（Li7500A）之间的距离是17cm。
涡动相关仪的原始观测数据为10Hz，发布的数据是采用Eddypro软件处理的30分钟数据，其处理的主要步骤包括：野点值剔除，延迟时间校正，坐标旋转（二次坐标旋转），频率响应修正，超声虚温修正和密度（WPL）修正等。同时对各通量值进行质量评价，主要是大气平稳性（Δst）和湍流相似性特征（ITC）的检验。对Eddypro软件输出的30min通量值也进行了筛选：（1）剔除仪器出错时的数据；（2）剔除降水前后1h的数据；（3）剔除10Hz原始数据中每30min内缺失率大于10%的数据。观测数据的平均周期为30分钟，一天48个数据，缺失数据标记为-6999。4月3日-4日涡动系统的Li7500A进行标定，导致数据缺失。
发布的观测数据包括：日期/时间Date/Time，风向Wdir（°），水平风速Wnd（m/s），侧向风速标准差Std_Uy（m/s），超声虚温Tv（℃），水汽密度H2O（g/m3），二氧化碳浓度CO2（mg/m3），摩擦速度Ustar（m/s），奥布霍夫长度L（m），感热通量Hs（W/m2），潜热通量LE（W/m2），二氧化碳通量Fc（mg/(m2s)），感热通量的质量标识QA_Hs，潜热通量的质量标识QA_LE，二氧化碳通量的质量标识QA_Fc。感热、潜热、二氧化碳通量的质量标识分为九级（质量标识1-3数据质量好，4-6数据质量较好，7-8数据质量较差（较插补数据好）；9数据质量差）。数据时间的含义，如0:30代表0:00-0:30的平均；数据以*.xls格式存储。
观测数据处理请参考Liu et al. (2011)。

This dataset contains eddy covariance observations collected at the Guazhou Station of the Cold and Arid Regions Science Observation Network of Lanzhou University, spanning from September 24 to December 31, 2018. The station is located in Liuyuan Town, Guazhou County, Jiuquan, Gansu Province, with a desert underlying surface. The observation site has coordinates of 95.673°E, 41.405°N and an elevation of 2014 m. The eddy covariance system has an installation height of 4 m, a sampling frequency of 10 Hz, and the ultrasonic anemometer is oriented towards true north. The distance between the ultrasonic sonic temperature anemometer (CSAT3) and the CO2/H2O gas analyzer (Li7500A) is 17 cm.

The raw observational data from the eddy covariance system is at 10 Hz. The released dataset consists of 30-minute averaged data processed using the EddyPro software. The main processing steps include: outlier removal, time lag correction, coordinate rotation (double coordinate rotation), frequency response correction, sonic virtual temperature correction, and density (WPL) correction, among others. Meanwhile, quality assessment was conducted for all flux values, primarily including tests of atmospheric stability (Δst) and turbulent similarity characteristics (ITC). Additional screening was applied to the 30-minute flux values output by EddyPro: (1) removal of data collected during instrument malfunctions; (2) removal of data within 1 hour before and after precipitation events; (3) removal of data with a missing rate exceeding 10% within each 30-minute window of the 10 Hz raw observations. The observational data are averaged over 30-minute intervals, resulting in 48 data points per day. Missing data are marked as -6999. Data were missing on April 3–4 due to calibration of the Li7500A analyzer in the eddy covariance system.

The released observational data include: Date/Time, wind direction Wdir (°), horizontal wind speed Wnd (m/s), lateral wind speed standard deviation Std_Uy (m/s), sonic virtual temperature Tv (℃), water vapor density H2O (g/m³), carbon dioxide concentration CO2 (mg/m³), friction velocity Ustar (m/s), Obukhov length L (m), sensible heat flux Hs (W/m²), latent heat flux LE (W/m²), carbon dioxide flux Fc (mg/(m²s)), quality flag of sensible heat flux QA_Hs, quality flag of latent heat flux QA_LE, and quality flag of carbon dioxide flux QA_Fc. The quality flags for sensible heat flux, latent heat flux, and carbon dioxide flux are divided into nine levels: quality flags 1–3 indicate good data quality, 4–6 indicate fair data quality, 7–8 indicate poor data quality (better than interpolated data), and 9 indicates very poor data quality. The meaning of the data timestamp is as follows: for example, 0:30 represents the average over the period 0:00–0:30. All data are stored in *.xls format.

Please refer to Liu et al. (2011) for details on the observational data processing procedures.