The dataset contains vegetation type in the middle reaches of the Heihe River Basin, which was used to validate products from remote sensing. It was generated from investigating the land cover strips of CASI during 2012. Instruments: High-precision handheld GPS (2-3 m) and digital camera were used as main tools in the survey. Measurement method: Hierarchical classification is applied based on CASI data. According to various land types, pixel classifications is used for forest, grassland, bare land and building lands; in-situ observations and investigations are used for different crops. Dataset contains: land types, including maize, leek, poplar trees, cauliflower, bell pepper, potatoes, endive sprout, orchard, watermelon, kidney bean, pear orchard, shadow, and non-vegetation, except for 14 others which are not classified. Observation site: core experimental areas with 5*5 matrix structure in the middle reaches of the Heihe river basin Date: From 25 June in 2012 (UTC+8) on.

The datasets of “Land Cover Map of Heihe River Basin” provide monthly land cover classification data in 2012-2013. The HJ-1/CCD data with both high spatial resolution (30 m) and high temporal (2 days) frequency was used to construct the time series data. The NDVI curves from the time series HJ-1/CCD data can depict the variation of typical land surface. Different land use type has different NDVI curve. Rules were set to extract every land use type information. The datasets of “Land Cover Map of Heihe River Basin” hold the traditional land use types including water bodies, urban and built-up, croplands, evergreen coniferous forests, deciduous broadleaf forests and so on. Crop type classification (including maize, spring wheat, highland barely, rape and so on), snow and ice and glaciers information updates, make the datasets more detailed. Compared with previous land cover map and other products, the classification result of the datasets is visually bette. Especially in middle stream, the accuracy of crop classification is quite high compared with the data from the ground campaign. The accuracy of land cover map of the datasets in 2012 was evaluated using very high spatial resolution remote sensing data within Google Earth and data from campaign, and the overall accuracy can be as high as 92.19%. In a word, the datasets of “Land Cover Map of Heihe River Basin” is not only high in overall accuracy, but also more detailed in crop fine classification. Furthermore, it updated some new classes like glaciers and snow. The datasets of “Land Cover Map of Heihe River Basin” are consequently the classification datasets with the highest accuracy and most detailed information up to now.

This dataset contains the automatic weather station (AWS) measurements from site No.14 in the flux observation matrix from 6 May to 21 September, 2012. The site (100.35310° E, 38.85867° N) was located in a cropland (maize surface) in Yingke irrigation district, which is near Zhangye, Gansu Province. The elevation is 1570.23 m. The installation heights and orientations of different sensors and measured quantities were as follows: air temperature and humidity (HMP45D; 5 m, towards north), air pressure (CS100; 2 m), rain gauge (TE525M; 10 m), wind speed and direction (034B; 10 m, towards north), a four-component radiometer (CNR4; 6 m, towards south), two infrared temperature sensors (IRTC3; 6 m, vertically downward), soil temperature profile (AV-10T; 0, -0.02, -0.04, -0.1, -0.2, -0.4, -0.6, and -1.0 m), soil moisture profile (ECh2o-5; -0.02, -0.04, -0.1, -0.2, -0.4, -0.6, and -1.0 m), and soil heat flux (HFT3; 3 duplicates with one below the vegetation and the other between plants, 0.06 m). The observations included the following: air temperature and humidity (Ta_5 m and RH_5 m) (℃ and %, respectively), air pressure (press, hpa), precipitation (rain, mm), wind speed (Ws_10 m, m/s), wind direction (WD_10 m, °), 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 IR_2, ℃), soil heat flux (Gs_1, below the vegetation; Gs_2 and Gs_3, W/m^2), soil temperature profile (Ts_0 cm, Ts_2 cm, Ts_4 cm, Ts_10 cm, Ts_20 cm, Ts_40 cm, Ts_60 cm, and Ts_100 cm, ℃), and soil moisture profile (Ms_2 cm, Ms_4 cm, Ms_10 cm, Ms_20 cm, Ms_40 cm, Ms_60 cm, and Ms_100 cm, %). The data processing and quality control steps were as follows. (1) The AWS data were averaged over intervals of 10 min; therefore, there were 144 records per day. The missing data were filled with -6999. (2) Data in duplicate records were rejected. (3) Unphysical data were rejected. (4) In this dataset, the time of 0:10 corresponds to the average data for the period between 0:00 and 0:10; the data were stored in *.xlsx format. (5) Finally, the naming convention was AWS+ site no. Moreover, suspicious data were marked in red. For more information, please refer to Liu et al. (2016) (for multi-scale observation experiment or sites information), Xu et al. (2013) (for data processing) in the Citation section.

On 30 June 2012 (UTC+8), TASI sensor carried by the Harbin Y-12 aircraft was used in a visible near Infrared hyperspectral airborne remote sensing experiment, which is located in the observation experimental area (30×30 km), Linze region and Heihe riverway. The relative flight altitude is 2500 meters. The wavelength of TASI is 8-11.5 μm with a spatial resolution of 3 meters. Through the ground sample points and atmospheric data, the data are recorded in surface radiance processed by geometric correction and atmospheric correction. Land surface temperature (LST) data was retrieved by temperature/emissivity separation algorithm.

This dataset includes the BRF observations of the corn in the Daman site (100.372° E, 38.855° N) on 29-6-2012) and the desert site around the airport (100.700° E, 38.762° N) acquired on 8-7-2012. Instruments: SVC-HR1024 from IRSA, reference board from IRSA, the multi-angular auto-observing shelf developed by BNU Measurement methods: we measure the BRF in the unit of observing plane, i.e. fix the view azimuth then change the view zenith angle to measure the target spectra, including along the principle plane and cross the principle plane at different sun angle. Besides, the planes along and cross the ridge of corn are also measured, specific planes like 0° , 90° away from the north are also observed in the desert. In each observing plane, view zenith angles from -60° to 60° with a interval of 10° are observed. The fiber optic probe with a view field of 25° is fixed at the multi-angular shelf at a height of 5 meters. The spectrum measured by the SVC-HR1024 is ranged from 350 nm-2500 nm. In each plane measurement , the spectral radiance of the reference board is measured first, then the target radiance of different view zenith angle is measured, finally the reference board radiance is measured again. Dataset contains the originally recorded data like the spectra (in sig format) and the log files (in txt format), and the processed data BRDF (in txt format and jpg format). The processed data in the format of txt, contains the observing geometries and corresponding reflectance spectra from 350 nm to 2500 nm. The processed data in the format of jpg, is a quick view of the BRF at 550 nm, 650 nm and 850 nm of each observing plane.

On 25 July 2012, a Leica ALS70 airborne laser scanner boarded on the Y-12 aircraft was used to obtain the point cloud data. Leica ALS70 airborne laser scanner has unlimited numbers of returns intensities measurements including the first, second, third return intensities. The wavelength of laser light is 1064 nm. The absolute flight altitude is 4800 m with the point cloud density 1 points per square meter. Aerial LiDAR-DEM was obtained through parameter calibration, automatic classification of point cloud density and manual editing.

This dataset includes data recorded by the Hydrometeorological observation network obtained from the automatic weather station (AWS) at the observation system of Meteorological elements gradient of Shenshawo sandy desert station between 1 September, 2012, and 31 December, 2013. The site (100.493° E, 38.789° N) was located on a desert surface in the Shenshawo, which is near Zhangye city, Gansu Province. The elevation is 1594 m. The installation heights and orientations of different sensors and measured quantities were as follows: air temperature and humidity profile (HMP45AC; 5 and 10 m, north), wind speed profile (010C; 5 and 10 m, north), wind direction profile (020C; 10 m, north), air pressure (PTB110; 2 m), rain gauge (52203; 10 m), four-component radiometer (CNR1; 6 m, south), two infrared temperature sensors (IRTC3; 6 m, south, vertically downward), soil heat flux (HFP01; 3 duplicates, -0.06 m), soil temperature profile (109; 0, -0.02, -0.04, -0.1, -0.2, -0.4, -0.6, and -1 m), and soil moisture profile (CS616; -0.02, -0.04, -0.1, -0.2, -0.4, -0.6 and -1 m). The observations included the following: air temperature and humidity (Ta_5 m and Ta_10 m; RH_5 m and RH_10 m) (℃ and %, respectively), wind speed (Ws_5 m and Ws_10 m) (m/s), wind direction (WD_10 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_0 cm, Ts_2 cm, Ts_4 cm, Ts_10 cm, Ts_20 cm, Ts_40 cm, Ts_60 cm and Ts_100 cm) (℃), and soil moisture (Ms_2 cm, Ms_4 cm, Ms_10 cm, Ms_20 cm, Ms_40 cm, Ms_60 cm and Ms_100 cm) (%, volumetric water content). 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 precipitation data were missing during 31 March, 2013 and 26 July, 2013 because of the malfunction of rain gauge. 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: 2013-6-10 10:30. (6) Finally, the naming convention was AWS+ site no. Moreover, suspicious data were marked in red. For more information, please refer to Li et al. (2013) (for hydrometeorological observation network or sites information), Liu et al. (2011) (for data processing) in the Citation section.

The dataset of airborne WiDAS mission was obtained in the Linze station-Linze grassland flight zone on Jul. 11, 2008. Intra-band data available for general users include Level-2C data (after geometric, radiometric and atmospheric corrections), Level-1B browse image (after intra-band matching) and Level-2B browse image (after registration). The raw data, Level-1A, and data processing parameters were filed; applications would be evaluated prior to access. Data processing started in Aug. 2008 and ended in Apr. 2009, and in Nov. 2009, CCD data were reprocessed to adjust radiometric calibration. The flying time of each route was as follows: {| ! id ! flight ! relative height ! starttime ! endtime ! data size ! data state ! data quality ! ground targets |- | 1 || 1#13 || 1500m || 1:52:06 || 11:58:02 || 90 || processed; compelete || good || Pingchuan reservoir |- | 2 || 1#11 || 1500m || 12:11:38 || 12:09:54 || 95 || processed; compelete || good || Linze grassland station |- | 3 || 1#9 || 1500m || 12:14:58 || 12:20:42 || 87 || processed; compelete || good || Pingchuan reservoir |- | 4 || 1#7 || 1500m || 12:27:14 || 12:33:18 || 92 || processed; compelete || good || desert transit zone |- | 5 || 1#5 || 1500m || 12:38:22 || 12:44:14 || 89 || processed; compelete || good || north-south desert plot |- | 6 || 1#3 || 1500m || 12:50:30 || 12:56:26 || 90 || processed; compelete || good || Pingchuan reservoir |- | 7 || 1#1 || 1500m || 13:01:46 || 13:07:46 || 91 || processed; compelete || good || Linze station |}

The data set includes the estimated data on the SOS (start of season) and the EOS (end of season) of vegetation in Sanjiangyuan based on GIMMS3g version 1.0, the latest version of the GIMMS NDVI data set. Two common phenological estimation methods were adopted: the threshold extraction method based on polynomial fitting (the term “poly” was included in the file names) and the inflection point extraction method based on double logistic function fitting (the term “sig” was included in the file names). These data can be used to analyse the relationship between vegetation phenology and climate change. The temporal coverage ranges from 1982 to 2015, and the spatial resolution is 8 km.

This dataset includes data recorded by the Hydrometeorological observation network obtained from an observation system of Meteorological elements gradient of Sidaoqiao Superstation between 11 July, 2013, and 31 December, 2013. The site (101.137° E, 42.001° N) was located on a tamarix (Tamarix chinensis Lour.) surface in the Sidaoqiao, Dalaihubu Town, Ejin Banner, Inner Mongolia Autonomous Region. The elevation is 873 m. The installation heights and orientations of different sensors and measured quantities were as follows: air temperature and humidity profile (HC2S3; 5, 7, 10, 15, 20 and 28 m, towards north), wind speed profile (010C; 5, 7, 10, 15, 20 and 28 m, towards north), wind direction profile (020C; 15 m, towards north), air pressure (CS100; in waterproof box), rain gauge (TE525M; 28 m, towards south), four-component radiometer (CNR4; 10 m, towards south), two infrared temperature sensors (SI-111; 10 m, towards south, vertically downward), two photosynthetically active radiation (PQS-1; 10 m, towards south, one vertically upward and one vertically downward), soil heat flux (HFP01SC; 3 duplicates with G1 below the tamarix; G2 and G3 between plants, -0.06 m), a TCAV averaging soil thermocouple probe (installed on 17 July, 2013, TCAV; -0.02, -0.04 m), soil temperature profile (109ss-L; 0, -0.02, -0.04, -0.1, -0.2, -0.4, -0.8, -1.2 and -1.6 m), and soil moisture profile (install on 7 December, 2013, ML2X; -0.02, -0.04, -0.1, -0.2, -0.4, -0.8, -1.2 and -1.6 m). The observations included the following: air temperature and humidity (Ta_5 m, Ta_7 m, Ta_10 m, Ta_15 m, Ta_20 m and Ta_28 m; RH_5 m, RH_7 m, RH_10 m, RH_15 m, RH_20 m and RH_28 m) (℃ and %, respectively), wind speed (Ws_5 m, Ws_7 m, Ws_10 m, Ws_15 m, Ws_20 m and Ws_28 m) (m/s), wind direction (WD_15 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) (℃), photosynthetically active radiation of upward and downward (PAR_up and PAR_down) (μmol/ (s m^-2)), average soil temperature (TCAV, ℃), soil heat flux (Gs_1, Gs_2 and Gs_3) (W/m^2), soil temperature (Ts_0 cm, Ts_2 cm, Ts_4 cm, Ts_10 cm, Ts_20 cm, Ts_40 cm, Ts_80 cm, Ts_120 cm and Ts_160 cm) (℃), and soil moisture (Ms_2 cm, Ms_4 cm, Ms_10 cm, Ms_20 cm, Ms_40 cm, Ms_80 cm, Ms_120 cm and Ms_160 cm) (%, volumetric water content). 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 wind speed (10 m height) data were missing before 12 November, 2013 because of the sensor problem. 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: 2013-9-10 10:30. (6) Finally, the naming convention was AWS+ site no. Moreover, suspicious data were marked in red. For more information, please refer to Li et al. (2013) (for hydrometeorological observation network or sites information), Liu et al. (2011) (for data processing) in the Citation section.

In the mid-latitude region of Asia, the southeastern region is humid and affected by monsoon circulation (thus, it is referred to as the monsoon region), and the inland region is arid and controlled by the other circulation patterns (these areas include the cold and arid regions in the northern Tibetan Plateau, referred to as the westerly region). Based on the generalization of the climate change records published in recent years, the westerly region was humid in the mid-late Holocene, which was significantly different from the pattern of the Asian monsoon in the early-middle Holocene. In the past few millennia, the westerly region was arid during the Medieval Warm Period but relatively humid during the Little Ice Age. In contrast, the oxygen isotope records derived from a stalagmite in the Wanxiang Karst Cave showed that the monsoon precipitation was high in the Medieval Warm Period and low during the Little Ice Age. In the last century, especially in the last 50 years, the humidity of the arid regions in the northwest has increased, while the eastern areas of northwestern and northern China affected by the monsoon have become more arid. Moreover, in the northern and southern parts of the Tibetan Plateau, which are affected by the westerlies and the monsoon, respectively, the precipitation changes on the interdecadal and century scales have also shown an inverse phase. Based on these findings, we propose that the control zone of the westerly belt in central Asia has different humidity (precipitation) variation patterns than the monsoon region on every time scale (from millennial to interdecadal) in the modern interglacial period. The integrated research project on Holocene climate change in the arid and semi-arid regions of western China was a major research component of the project Environmental and Ecological Science for West China, which was funded by the National Natural Science Foundation of China. The leading executive of the project was Professor Fahu Chen from Lanzhou University. The project ran from January 2006 to December 2009. The data collected by the project include the following: 1. The integrate humidity data over the Holocene in the arid regions of Central-East Asia and 12 lakes (11000-0 cal yr BP): including Lake Van, Aral Sea, Issyk-Kul, Ulunguhai Lake, Bosten Lake, Barkol Lake, Bayan Nuur, Telmen Lake, Hovsgol Nuur, Juyan Lake, Gun Nuur and Hulun Nuur. 2. The integrated humidity data over the past millennium in the arid regions of Central-East Asia and at five research sites (1000-2000): including Aral Sea, Guliya, Bosten Lake, Sugan Lake, and the Badain Juran desert. Data format: excel table.

This dataset includes data recorded by the Hydrometeorological observation network obtained from the automatic weather station (AWS) at the observation system of Meteorological elements gradient of Sidaoqiao barren-land station between 9 July, 2013, and 31 December, 2013. The site (101.133° E, 41.999° N) was located on a barren-land surface in the Sidaoqiao, Dalaihubu Town, Ejin Banner, Inner Mongolia Autonomous Region. The elevation is 878 m. The installation heights and orientations of different sensors and measured quantities were as follows: four-component radiometer (CNR4; 24 m, south), two infrared temperature sensors (SI-111; 24 m, south, vertically downward), soil heat flux (HFP01; 3 duplicates, -0.06 m), and soil temperature profile (AV-10T; 0, -0.02 and -0.04 m). The observations included the following: 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), and soil temperature (Ts_0 cm, Ts_2 cm, Ts_4 cm) (℃). 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. Data were missing during 24 September, 2013 and 26 September, 2013 because of the malfunction of datalogger. 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: 2013-9-10 10:30. (6) Finally, the naming convention was AWS+ site no. Moreover, suspicious data were marked in red. For more information, please refer to Li et al. (2013) (for hydrometeorological observation network or sites information), Liu et al. (2011) (for data processing) in the Citation section.