A decadal (2008-2017) daily evapotranspiration data set of 1 km spatial resolution and spatial completeness across the North China Plain using TSEB and data fusion

Authors: ZHANG, CAIJIN - Tsinghua University; LONG, D.S. - Tsinghua University; ZHANG, Y. - Chinese Academy Of Sciences; Anderson, Martha; Kustas, William - Bill; YANG, YANG - US Department Of Agriculture (USDA)

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/16/2021
Publication Date: 5/23/2021
Citation: Zhang, C., Long, D., Zhang, Y., Anderson, M.C., Kustas, W.P., Yang, Y. 2021. A decadal (2008-2017) daily evapotranspiration data set of 1 km spatial resolution and spatial completeness across the North China Plain using TSEB and data fusion. Remote Sensing of Environment. 262:112519. https://doi.org/10.1016/j.rse.2021.112519.
DOI: https://doi.org/10.1016/j.rse.2021.112519

Interpretive Summary: The North China Plain (NCP) is the major crop production base in China with widespread and increasing irrigation, drawing water supply from canals and groundwater resources. The ability to monitor consumptive water use (evapotranspiration) routinely and with reasonable accuracy is critical to the long-term sustainability of the NCP production system. However, most current remote sensing approaches for monitoring evapotranspiration are affected by cloud cover, which limits the frequency with which spatial maps can be updated in this region. This paper discusses a means for time-continuous evapotranspiration mapping using gap-filled thermal infrared image products, a key model input providing proxy information about the surface moisture status. The all-sky modeling system is evaluated with ground measurements from several flux sites in the NCP, and derived spatiotemporal patterns of water use are commensurate with known patterns in climate and land-use. The model should be valuable for agricultural water resource management for the NCP, and has potential to be applied to similar agricultural regions globally.

Technical Abstract: Daily continuous evapotranspiration (ET) estimates of 1 km spatial resolution can benefit agricultural water resources management at regional scales. Thermal infrared remote sensing-derived land surface temperature (LST) is a critical variable for ET estimation using energy balance-based models. However, missing LST information under cloudy conditions remains a long-standing barrier for spatiotemporally continuous monitoring of daily ET at regional scales. In this study, LST data of 1 km spatial resolution at 11:00 local solar time under all-weather conditions across the North China Plain (NCP) were first generated using a data fusion approach. Second, combined with the generated LST data, MODIS products, and meteorological forcing from the China Land Data Assimilation System, the Two-Source Energy Balance model (TSEB) and a temporal upscaling method were jointly used to estimate daily ET at 1 km spatial resolution across the NCP for a decade from 2008'2017. To better incorporate the impact of crop phenology on ET and improve the ET estimation, the fraction of greenness in TSEB was determined in terms of a leaf area index threshold during the crop growth period, as opposed to be 1 in many published studies. Compared with observed instantaneous latent heat flux (LE) with energy balance closure, the estimated LE can well capture inter- and intra-annual variations, in terms of R² of 0.63 ' 0.79 at the Huailai, Daxing, Weishan, and Guantao flux towers. Compared with observed instantaneous LE without energy balance closure, R² of the LE estimates at six flux towers (i.e., Huailai, Daxing, Luancheng, Yucheng, Weishan, and Guantao) varied from 0.49 to 0.68. R² of the estimated daily ET with energy balance closure at the Huailai, Daxing, and Guantao sites was greater than 0.70, and the RMSE was below 0.88 mm/d; whereas compared with unclosed daily ET measurements at the five sites (i.e., Huailai, Daxing, Luancheng, Yucheng, and Guantao), R² of the estimated daily ET was greater than 0.51, and the RMSE was about ~1 mm/d. These accuracies are comparable with published results, with our ET data set featuring spatiotemporal continuity and high spatial resolution across the entire NCP for a decade.