Core Ideas: Heihe was the first basin‐scale integrated observatory network established in China. An intensive flux observation matrix experiment was conducted. New techniques, e.g., wireless sensor network, flux matrix, and airborne remote sensing, are used. The integrated observatory network is useful in land surface processes research. Research on land surface processes at the catchment scale has drawn much attention over the past few decades, and a number of watershed observatories have been established worldwide. The Heihe River Basin (HRB), which contains the second largest inland river in China, is an ideal natural field experimental area for investigation of land surface processes involving diverse landscapes and the coexistence of cold and arid regions. The Heihe Integrated Observatory Network was established in 2007. For long‐term observations, a hydrometeorological observatory, ecohydrological wireless sensor network, and satellite remote sensing are now in operation. In 2012, a multiscale observation experiment on evapotranspiration over heterogeneous land surfaces was conducted in the midstream region of the HRB, which included a flux observation matrix, wireless sensor network, airborne remote sensing, and synchronized ground measurements. Under an open data policy, the datasets have been publicly released following careful data processing and quality control. The outcomes highlight the integrated research on land surface processes in the HRB and include observed trends, scaling methods, high spatiotemporal resolution remote sensing products, and model–data integration in the HRB, all of which are helpful to other endorheic basins in the “Silk Road Economic Belt.” Henceforth, the goal of the Heihe Integrated Observatory Network is to develop an intelligent monitoring system that incorporates ground‐based observatory networks, unmanned aerial vehicles, and multi‐source satellites through the Internet of Things technology. Furthermore, biogeochemical processes observation will be improved, and the study of integrating ground observations, remote sensing, and large‐scale models will be promoted further.

Agreement among instruments is very important for the Multi‐Scale Observation Experiment on Evapotranspiration over heterogeneous land surfaces of The Heihe Watershed Allied Telemetry Experimental Research (HiWATER‐MUSOEXE), particularly in regard to radiation and turbulent flux measurements. Before HiWATER‐MUSOEXE was conducted, 20 eddy covariance (EC) system sets, 18 radiometer sets, and seven large aperture scintillometers (LASs) sets were intercompared over the Gobi desert between 14 and 24 May 2012. For radiometers, the four‐component radiation measurements exhibited good agreement—the average root‐mean‐square error (RMSE) and mean relative error (MRE) for the net radiation were 10.38 W m⁻² and 1.24%, respectively. With regard to the EC systems, the best consistency for sensible heat fluxes was found among CSAT3 sonic anemometers and Li7500A/Li7500/EC150 combinations (average RMSE, 12.30 W m⁻² and MRE, −1.36%), followed by Gill sonic anemometers and Li7500A/Li7500 combinations when a proper angle of attack correction method was applied (average RMSE, 16.75 W m⁻² and MRE, −5.52%). The sensible heat flux measured using different LASs agreed well with high correlation coefficients—the average RMSE and MRE values were 10.26 W m⁻² and 5.48% for boundary layer scintillometer (BLS) 900, 16.32 W m⁻² and 10.47% for BLS450, and 14.38 W m⁻² and −3.72% for ZZLAS, respectively. The EC and LAS measurements were compared and agreed well over homogeneous underlying surfaces, which also indicated that the EC and LAS measurements would be comparable in the follow‐up experiment. The intercomparison results can be used to determine instrument placement and are very helpful for subsequent data analysis.

Evapotranspiration (ET) is a vital variable for land‐atmosphere interactions that links surface energy balance, water, and carbon cycles. The in situ techniques can measure ET accurately but the observations have limited spatial and temporal coverage. Modeling approaches have been used to estimate ET at broad spatial and temporal scales, while accurately simulating ET at regional scales remains a major challenge. In this study, we upscale ET from eddy covariance flux tower sites to the regional scale with machine learning algorithms. Five machine learning algorithms are employed for ET upscaling including artificial neural network, Cubist, deep belief network, random forest, and support vector machine. The machine learning methods are trained and tested at 36 flux towers sites (65 site years) across the Heihe River Basin and are then applied to estimate ET for each grid cell (1 km × 1 km) within the watershed and for each day over the period 2012–2016. The artificial neural network, Cubist, random forest, and support vector machine algorithms have almost identical performance in estimating ET and have slightly lower root‐mean‐square error than deep belief network at the site scale. The random forest algorithm has slightly lower relative uncertainty at the regional scale than other methods based on three‐cornered hat method. Additionally, the machine learning methods perform better over densely vegetated conditions than barren land or sparsely vegetated conditions. The regional ET generated from the machine learning approaches captured the spatial and temporal patterns of ET at the regional scale.