"Coupling and Evolution of Hydrologic -Ecologic-Economic Processes of the Heihe River Basin Under the Framework of Water Rights" (91125018) Project data collection 1 - SWater Resources Improvement Plan of Shiyang River Basin 1. Data Overview:The improvement plan of Shiyang River Basin was implemented in 2007 for river basin comparison. 2. Data Content: The released plan.

The land use / land cover data set of Heihe River Basin in 2011 is the Remote Sensing Research Office of Institute of cold and drought of Chinese Academy of Sciences. Based on the remote sensing data of landsatm and ETM in 2011, combined with field investigation and verification, a 1:100000 land use / land cover image and vector database of Heihe River Basin is established. The data set mainly includes 1:100000 land use graph data and attribute data in the middle reaches of Heihe River Basin. The land cover data of 1:100000 (2011) in Heihe River Basin and the previous land cover are classified into six first-class categories (cultivated land, forest land, grassland, water area, urban and rural residents, industrial and mining land and unused land) and 25 second-class categories by the same hierarchical land cover classification system. The data type is vector polygon and stored in shape format. Land cover classification attributes: Level 1 type level 2 type attribute code spatial distribution location Cultivated land: plain dry land 123 is mainly distributed in basin, piedmont, river alluvial, proluvial or lacustrine plain (poor irrigation conditions due to water shortage). The upland and land 122 is mainly distributed in the hilly area, and generally, the plot is distributed on the gentle slope of the hill, as well as on the top of the ridge and the base. The dry land 121 is mainly distributed in the mountainous area, the hillside (gentle slope, hillside, steep slope platform, etc.) and the Piedmont belt below 4000 m above sea level. Woodland: there are woodland (Arbor) 21 mainly distributed in high mountains (below 4000 meters above sea level) or middle mountain slopes, valley slopes, mountain tops, plains, etc. Shrub land 22 is mainly distributed in the higher mountain area (below 4500m), most of which are hillside, valley and sandy land. Sparse forest land 23 is mainly distributed in mountainous areas, hills, plains and sandy land, Gobi (Loamy, sandy conglomerate) edge. Other forest lands 24 are mainly distributed around the oasis ridge, riverside, roadside and rural residential areas. Grassland: high cover grassland 31 is generally distributed in mountainous area (gentle slope), hilly area (steep slope), river beach, Gobi, sandy land, etc. The middle cover grassland 32 is mainly distributed in dry areas (low-lying land next door and land between Sandy Hills, etc.). Low cover grassland 33 mainly grows in dry areas (loess hills and sand edge). Water area: channel 41 is mainly distributed in plain, inter Sichuan cultivated land and inter mountain valley. Lake 42 is mainly distributed in low-lying areas. Reservoir pond 43 is mainly distributed in plain and valley between rivers, surrounded by residential land and cultivated land. Glaciers and permanent snow cover 44 are mainly distributed on the top of (over 4000) mountains. The beach land 46 is mainly distributed in the valley, piedmont, plain lowland, the edge of river lake basin and so on. Residential land: urban land 51 is mainly distributed in plain, mountain basin, slope and gully platform. Rural residential land 52 is mainly distributed in oasis, cultivated land and roadside, tableland, slope, etc. Industrial and mining land and traffic land 53 are generally distributed in the periphery of cities and towns, more developed traffic areas and industrial mining areas. Unused land: sand 61 is mostly distributed in the basin, both sides of the river, the river bay and the periphery of the mountain front Gobi. Gobi 62 is mainly distributed in the Piedmont belt with strong wind erosion and sediment transport. Salt alkali 63 is mainly distributed in relatively low and easy to accumulate water, dry lakes and lakeside. Swamp 64 is mainly distributed in relatively low and easy to accumulate water. Bare soil 65 is mainly distributed in the arid areas (mountain steep slopes, hills, Gobi), and the vegetation coverage is less than 5%. Bare rock 66 is mainly distributed in the extremely dry stone mountain area (windy, light rain). The other 67 are mainly distributed in the exposed rocks formed by freezing and thawing over 4000 meters, also known as alpine tundra. Projection parameters: Projection ALBERS Units METERS Spheroid Krasovsky Parameters: 25 00 0.000 /* 1st standard parallel 47 00 0.000 /* 2nd standard parallel 105 00 0.000 /* central meridian 0 0 0.000 /* latitude of projection's origin 0.00000 /* false easting (meters) 0.00000 /* false northing (meters)

This data set is extracted from the second Glacier Inventory Data Set of China for Three River Source area. The file is SHP format. The attribute data are as follows: Glc_Name (glacier name), Drng_Code (basin code), FCGI_ID (first glacier catalogue code), GLIMS_ID (GLIMS glacier code), Mtn_Name (mountain system name), Pref_Name (administrative division), Glc_Long (glacier longitude), Glc_Lati (glacier latitude), Glc_Area (glacier area), Abs_Accu (absolute area accuracy), Rel_Accu (relative area accuracy), Deb_Area (surface Moraine Area), Deb_A_Accu (absolute accuracy of surface moraine Area), Deb_R_Accu (relative accuracy of surface moraine area)、Glc_Vol_A (estimation of glacier volume 1)、Glc_Vol_B (estimation of glacier volume 2)、Max_Elev (maximum glacier elevation)、Min_Elev (minimum glacier elevation)、Mean_Elev (average glacier elevation)、MA_Elev (median area height of glacier)、Mean_Slp (average glacier slope)、Mean_Asp (average glacier slope direction)、Prm_Image (major remote sensing data)、Aux_Image (auxiliary remote sensing data)、Rep_Date (glacier catalogue represents date)、Elev_Src (elevation data source)、Elev_Date (elevation represents date)、Compiler (glacier cataloguing editor)、Verifier (glacier cataloguing verifier).

This dataset contains the flux measurements from the Subalpine shrub eddy covariance system (EC) belonging to the Qinghai Lake basin integrated observatory network from April 28 to December 31 in 2019. The site (100°6'3.62"E, 37°31'15.67" N ) was located near Dasi, Shaliuhe Town, Gangcha County, Qinghai Province. The elevation is 3495m. The EC was installed at a height of 2.5m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (Gill&Li7500A) was about 0.17 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 3% of the 30 min raw record; and (4) data were rejected at night when the friction velocity (u*) was less than 0.1 m/s. There were 48 records per day, and the missing data were replaced with -6999. The released data contained the following variables: DATE/TIME, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). The quality marks of sensible heat flux, latent heat flux and carbon flux are divided into three levels (quality marks 0 have good data quality, 1 have good data quality and 2 have poor data quality). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references.

Wildfires can strongly affect the frozen soil environment by burning surface vegetation and soil organic matter. Vegetation affected by fire can take many years to return to mature pre-fire levels. In this data set, the effects of fires on vegetation regrowth in a frozen-ground tundra environment in the Anaktuvuk River Basin on the North Slope of Alaska were studied by quantifying changes in C-band and L-band SAR backscatter data over 15 years (2002-2017). After the fire, the C- and L-band backscattering coefficients increased by 5.5 and 4.4 dB, respectively, in the severe fire area compared to the unburned area. Five years after the fire, the difference in C-band backscattering between the fire zone and the unburned zone decreased, indicating that the post-fire vegetation level had recovered to the level of the unburned zone. This long recovery time is longer than the 3-year recovery estimated from visible wavelength-based NDVI observations. In addition, after 10 years of vegetation recovery, the backscattering of the L-band in the severe fire zone remains approximately 2 dB higher than that of the unburned zone. This continued difference may be caused by an increase in surface roughness. Our analysis shows that long-term SAR backscattering data sets can quantify vegetation recovery after fire in an Arctic tundra environment and can also be used to supplement visible-wavelength observations. The temporal coverage of the backscattering data is from 2002 to 2017, with a time resolution of one month, and the data cover the Anaktuvuk River area on the North Slope of Alaska. The spatial resolution is 30~100 m, the C- and L-band data are separated, and a GeoTIFF file is stored every month. For details on the data, see SAR Backscattering Data of the Anaktuvuk River Basin on the North Slope of Alaska - Data Description.

This dataset includes data recorded by the Qinghai Lake integrated observatory network obtained from an observation system of Meteorological elements gradient of the Alpine meadow and grassland ecosystem Superstation from August 31 to December 24, 2018. The site (98°35′41.62″E, 37°42′11.47″N) was located in the alpine meadow and alpine grassland ecosystem, near the SuGe Road in Tianjun County, Qinghai Province. The elevation is 3718m. The installation heights and orientations of different sensors and measured quantities were as follows: air temperature and humidity profile (HMP155; 3, 5, 10, 15, 20, 30, and 40 m, towards north), wind speed and direction profile (windsonic; 3, 5, 10, 15, 20, 30, and 40 m, towards north), air pressure (PTB110; 3 m), rain gauge (TE525M; 10m of the platform in west by north of tower), four-component radiometer (CNR4; 6m, towards south), two infrared temperature sensors (SI-111; 6 m, towards south, vertically downward), photosynthetically active radiation (PQS1; 6 m, towards south, each with one vertically downward and one vertically upward, soil heat flux (HFP01; 3 duplicates below the vegetation; -0.06 m), soil temperature profile (109; -0.05、-0.10、-0.20、-0.40、-0.80、-1.20、-2.00、-3.00 and -4.00m), soil moisture profile (CS616; -0.05、-0.10、-0.20、-0.40、-0.80、-1.20、-2.00、-3.00 and -4.00m). The observations included the following: air temperature and humidity (Ta_3 m, Ta_5 m, Ta_10 m, Ta_15 m, Ta_20 m, Ta_30 m, and Ta_40 m; RH_3 m, RH_5 m, RH_10 m, RH_15 m, RH_20 m, RH_30 m, and RH_40 m) (℃ and %, respectively), wind speed (Ws_3 m, Ws_5 m, Ws_10 m, Ws_15 m, Ws_20 m, Ws_30 m, and Ws_40 m) (m/s), wind direction (WD_3 m, WD_5 m, WD_10 m, WD_15 m, WD_20 m, WD_30m, and WD_40 m) (°), air pressure (press) (hpa), precipitation (rain) (mm), four-component radiation (DR, incoming shortwave radiation; UR, outgoing shortwave radiation; DLR_Cor, incoming longwave radiation; ULR_Cor, outgoing longwave radiation; Rn, net radiation) (W/m^2), infrared temperature (IRT_1 and IRT_2) (℃), soil heat flux (Gs_1, Gs_2, and Gs_3) (W/m^2), soil temperature (Ts_5cm、Ts_10cm、Ts_20cm、Ts_40cm、Ts_80cm、Ts_120cm、Ts_200cm、Ts_300cm、Ts_400cm) (℃), soil moisture (Ms_5cm、Ms_10cm、Ms_20cm、Ms_40cm、Ms_80cm、Ms_120cm、Ms_200cm、Ms_300cm、Ms_400cm) (%, volumetric water content), photosynthetically active radiation of upward and downward (PAR_D_up and PAR_D_down) (μmol/ (s m-2)). The data processing and quality control steps were as follows: (1) The AWS data were averaged over intervals of 10 min for a total of 144 records per day. The missing data were denoted by -6999. (2) Data in duplicate records were rejected. (3) Unphysical data were rejected. (4) The data marked in red are problematic data. (5) The format of the date and time was unified, and the date and time were collected in the same column, for example, date and time: 2018/8/31 10:30. Moreover, suspicious data were marked in red.

This dataset contains the flux measurements from the Qinghai Lake eddy covariance system (EC) belonging to the Qinghai Lake basin integrated observatory network from January 2 to October 18 in 2018. The site (100° 29' 59.726'' E, 36° 35' 27.337'' N) was located on the Yulei Platform in Erlangjian scenic area, Qinghai Province. The elevation is 3209m. The EC was installed at a height of 16.1m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (Gill&Li7500A) was about 0.17 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 3% of the 30 min raw record; and (4) data were rejected at night when the friction velocity (u*) was less than 0.1 m/s. There were 48 records per day, and the missing data were replaced with -6999. Data during October 13 to December 31, 2018 were absent due to the unavailable collecting condition in winter. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references.

This dataset contains the flux measurements from the Alpine meadow and grassland ecosystem Superstation superstation eddy covariance system (EC) belonging to the Qinghai Lake basin integrated observatory network from September 2 to December 18 in 2018. The site (98°35′41.62″E, 37°42′11.47″N) was located in the alpine meadow and alpine grassland ecosystem, near the SuGe Road in Tianjun County, Qinghai Province. The elevation is 3718m. The EC was installed at a height of 4.5 m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (CSAT3A &EC150) was about 0.17 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 3% of the 30 min raw record; and (4) data were rejected at night when the friction velocity (u*) was less than 0.1 m/s. There were 48 records per day, and the missing data were replaced with -6999. Data during December 18 to December 24, 2018 were missing due to the data collector failure. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references.

This dataset contains the flux measurements from the Guazhou station eddy covariance system (EC) in the middle reaches of the Heihe integrated observatory network from September 24 to December 31 in 2018. The site (95.673E, 41.405N) was located in a desert in Liuyuan Guazhou, which is near Jiuquan city in Gansu Province. The elevation is 2016 m. The EC was installed at a height of 4.0 m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (CSAT3&Li7500A) was 0.17 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 3% of the 30 min raw record; and (4) data were rejected at night when the friction velocity (u*) was less than 0.1 m/s. There were 48 records per day, and the missing data were replaced with -6999. Suspicious data were marked in red. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references. For more information, please refer to Liu et al. (2011) for data processing) in the Citation section.

This dataset contains the flux measurements from the Dashalong station eddy covariance system (EC) in the upperstream reaches of the Heihe integrated observatory network from January 1 to December 31 in 2018. The site (98.9406° E, 38.8399° N) was located in the Qilian County in Qinghai Province. The elevation is 3739 m. The EC was installed at a height of 4.5 m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (CSAT3&Li7500RS) was 0.15 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 3% of the 30 min raw record; and (4) data were rejected at night when the friction velocity (u*) was less than 0.1 m/s. There were 48 records per day, and the missing data were replaced with -6999. Suspicious data were marked in red. Data during September 27 to November 14, 2018 were missing due to the sensor calibration of sonic anemometer. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references. For more information, please refer to Liu et al. (2018) (for sites information), Liu et al. (2011) for data processing) in the Citation section.

This dataset contains the flux measurements from the Huazhaizi station eddy covariance system (EC) in the midstream reaches of the Heihe integrated observatory network from January 1 to December 31 in 2018. The site (100.3201° E, 38.7659° N) was located in the Zhangye City in Gansu Province. The elevation is 1731 m. The EC was installed at a height of 4.5 m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (CSAT3&Li7500A) was 0.15 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 10% of the 30 min raw record. There were 48 records per day, and the missing data were replaced with -6999. Suspicious data were marked in red. Data during May 13 to July 12 and July 16 to August 21, 2018 were missing due to the malfunction of CO2/H2O gas analyzer. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (z/L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references. For more information, please refer to Liu et al. (2018) (for sites information), Liu et al. (2011) for data processing) in the Citation section.

This dataset contains the flux measurements from the desert station eddy covariance system (EC) in the downstream reaches of the Heihe integrated observatory network from January 1 to December 31 in 2018. The site (100.9872° E, 42.1135° N) was located in the Ejina Banner in Inner Mongolia Autonomous Region. The elevation is 1054 m. The EC was installed at a height of 4.7 m, and the sampling rate was 10 Hz. The sonic anemometer faced north, and the separation distance between the sonic anemometer and the CO2/H2O gas analyzer (CSAT3&Li7500) was 0.15 m. The raw data acquired at 10 Hz were processed using the Eddypro post-processing software, including the spike detection, lag correction of H2O/CO2 relative to the vertical wind component, sonic virtual temperature correction, coordinate rotation (2-D rotation), corrections for density fluctuation (Webb-Pearman-Leuning correction), and frequency response correction. The EC data were subsequently averaged over 30 min periods. The observation data quality was divided into three classes according to the quality assessment method of stationarity (Δst) and the integral turbulent characteristics test (ITC): class 1-3 (high quality), class 4-6 (good), class 7-8 (poor, better than gap filling data), class9 (rejected). In addition to the above processing steps, the half-hourly flux data were screened in a four-step procedure: (1) data from periods of sensor malfunction were rejected; (2) data collected before or after 1 h of precipitation were rejected; (3) incomplete 30 min data were rejected when the missing data constituted more than 10% of the 30 min raw record. There were 48 records per day, and the missing data were replaced with -6999. Suspicious data were marked in red. Data during May 14 to June 26, 2018 were missing due to the data logger malfunction. The released data contained the following variables: data/time, wind direction (Wdir, °), wind speed (Wnd, m/s), the standard deviation of the lateral wind (Std_Uy, m/s), virtual temperature (Tv, ℃), H2O mass density (H2O, g/m3), CO2 mass density (CO2, mg/m3), friction velocity (ustar, m/s), stability (L), sensible heat flux (Hs, W/m2), latent heat flux (LE, W/m2), carbon dioxide flux (Fc, mg/ (m2s)), quality assessment of the sensible heat flux (QA_Hs), quality assessment of the latent heat flux (QA_LE), and quality assessment of the carbon flux (QA_Fc). In this dataset, the time of 0:30 corresponds to the average data for the period between 0:00 and 0:30; the data were stored in *.xls format. Detailed information can be found in the suggested references. For more information, please refer to Liu et al. (2018) (for sites information), Liu et al. (2011) for data processing) in the Citation section.