Yan et al. (2025) Multi-scale partitioning of evapotranspiration using the three-temperature model in an oasis–desert landscape of northwest China

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-10-06
• Authors: Chunhua Yan, Yongqiang Wang, Yujiu Xiong, Guo Yu Qiu
• DOI: 10.1016/j.ejrh.2025.102831

Research Groups

• Shenzhen Campus of Sun Yat-sen University, China
• School of Ecology, Sun Yat-sen University, Guangzhou, China
• School of Environment and Energy, Peking University Shenzhen Graduate School, Shenzhen, China
• School of Civil Engineering, Sun Yat-Sen University, Zhuhai, China
• Center of Water Resources and Environment, Sun Yat-sen University, Guangzhou, China

Short Summary

This study applied the three-temperature (3T) model to partition evapotranspiration (ET) into soil evaporation (LE) and plant transpiration (LT) across an oasis–desert landscape in Northwest China using multi-scale remote sensing data, demonstrating the model's consistent performance and scale resilience for areal mean ET partitioning from 3 meters to 1000 meters resolution.

Objective

• Validate the applicability of the three-temperature (3T) model for evapotranspiration (ET) partitioning across different spatial scales using isotope-based measurements.
• Quantify the influence of sensor resolution on the partitioning of transpiration (LT) and soil evaporation (LE).
• Assess the effects of spatial aggregation on the accuracy of ET partitioning.

Study Configuration

• Spatial Scale: The study was conducted in the middle reach of the Heihe River Basin, Hexi Corridor, Gansu Province, Northwest China, encompassing an oasis–desert landscape. Remote sensing data were analyzed at three resolutions: aerial (3 meters), ASTER (90 meters), and MODIS (1000 meters).
• Temporal Scale: Meteorological data were recorded from June to September 2012. Isotope-based measurements and remote sensing observations were primarily conducted on specific days during the growing season (e.g., July 10, August 2, 11, and 27, 2012), typically around midday (12:00–12:30 Beijing Time).

Methodology and Data

• Models used:
  • Three-temperature (3T) model for estimating and partitioning evapotranspiration (ET) into soil evaporation (LE) and plant transpiration (LT).
  • Craig–Gordon model for calculating midday transpiration fractions (T/ET) from stable oxygen and hydrogen isotope data.
  • Empirical equations for calculating fractional vegetation cover (f) from Normalized Difference Vegetation Index (NDVI), net radiation (Rn), atmospheric emissivity (εa), land surface albedo (α), and soil heat flux (G).
• Data sources:
  • Remote Sensing: Aerial imagery (3 meters resolution, visible, near-infrared, and thermal infrared) from the Heihe Watershed Applied Telemetry Experimental Research (HiWATER) project; ASTER thermal remote sensing data (90 meters resolution); MODIS thermal remote sensing data (1000 meters resolution).
  • In-situ Observations: Meteorological data (precipitation, air temperature, humidity, wind speed and direction, air pressure, solar radiation, net radiation) from a network of 21 automatic weather stations (10-minute intervals); stable oxygen and hydrogen isotope measurements (δ18O and δD in xylem water, soil water, and atmospheric vapor) using a tunable diode laser analyzer at the Daman Irrigation District; reference soil and vegetation surface temperatures measured with an IR Flexcam Ti55 Thermal Imager.
  • Derived Data: Normalized Difference Vegetation Index (NDVI) and Leaf Area Index (LAI).

Main Results

• The 3T model demonstrated good agreement with isotope-based measurements for the transpiration-to-evapotranspiration ratio (T/ET) in oasis croplands, with a mean absolute percentage error (MAPE) of 3.0 %, which was lower than the isotope uncertainty (5.2 %).
• The 3T model exhibited consistent performance in partitioning ET across three distinct spatial resolutions (3 meters, 90 meters, and 1000 meters).
• Finer-resolution data captured greater spatial variability in soil evaporation (LE) and plant transpiration (LT), showing higher maximum values and standard deviations, while mean LE and LT values remained relatively stable across scales.
• Strong consistency was observed between aggregated high-resolution and native coarse-resolution images for LE and LT, with coefficients of determination (R²) ranging from 0.59 to 0.88 and mean absolute errors (MAE) less than 50 W m⁻².
• Seasonal variations in LE and LT aligned with vegetation growth stages, with LT increasing and LE decreasing during rapid growth, and vice versa during decline.
• The 3T model accurately reproduced the spatial distribution of T/ET compared to the Two-Source Energy Balance (TSEB) model, with Pearson correlation coefficients of 0.87–0.88 and MAEs of -15 % to -18 %.

Contributions

• Systematically validated the three-temperature (3T) model for evapotranspiration (ET) partitioning using isotope-based measurements across a wide range of spatial scales (3 meters, 90 meters, 1000 meters) in a heterogeneous oasis-desert landscape.
• Demonstrated the 3T model's consistent performance and scale resilience for areal mean ET partitioning, highlighting its robustness for multi-scale applications.
• Quantified the impact of sensor resolution on the spatial variability and mean values of soil evaporation (LE) and plant transpiration (LT), providing insights into scale-dependent behavior.
• Confirmed the potential of the 3T model for reliable and scalable ET partitioning using moderate- to low-resolution satellite data, which is crucial for regional and global hydrological assessments, especially in data-scarce regions.

Funding

• Science, Technology and Innovation Commission of Shenzhen Municipality (Grant No. 202206193000001, 20220816152743007)
• National Natural Science Foundation of China (Grant No. 42071395 and 91025008)
• Heihe Watershed Applied Telemetry Experimental Research (HiWATER) project

Citation

@article{Yan2025Multiscale,
  author = {Yan, Chunhua and Wang, Yongqiang and Xiong, Yujiu and Qiu, Guo Yu},
  title = {Multi-scale partitioning of evapotranspiration using the three-temperature model in an oasis–desert landscape of northwest China},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.102831},
  url = {https://doi.org/10.1016/j.ejrh.2025.102831}
}