Refined assessment of potential evapotranspiration in the tarim basin

Wen Huo, Xiefei Zhi, Shunqi Hu, Wenyue Cai, Fan Yang, Chenglong Zhou, Ali Mamtimin, Qing He, Honglin Pan, Meiqi Song, Cong Wen, Yu Wang, Xinghua Yang, Lu MengFrontiers in Earth Science 2022 · DOI

TL;DR

This study refines the calculation of potential evapotranspiration (PET) in the Tarim Basin by comparing natural desert sands with human-made green belts using a modified Penman-Monteith-H (PM-H) model. The researchers found that human-engineered "green lands" significantly lower local PET compared to open desert, and they established a mean annual PET of 2,750 mm for the entire basin.

What Problem Does This Paper Address?

Accurately measuring potential evapotranspiration (PET)—the amount of water that could evaporate and transpire from a surface if water were unlimited—is critical for managing water resources in arid regions. However, the Taklimakan Desert is no longer just a sea of sand; human activities have introduced "Constructed Green Lands" (CGL) to combat desertification.

These green belts change the surface's albedo, roughness, and humidity, making standard PET models inaccurate. There is a significant gap in understanding how these human-altered surfaces differ from "Natural Sand Land" (NSL) in their hydrothermal processes. This paper seeks to refine PET assessments to account for these complex, patchy landscapes.

Methods

The authors employed the Penman-Monteith-H (PM-H) numerical model, an evolutionary version of the standard Penman-Monteith equation.

  • Refinement: Unlike standard models that rely on empirical constants, the PM-H model uses high-frequency "encrypted" observations to replace fixed values, allowing it to adapt to different underlying surfaces (sand vs. vegetation).
  • Sensitivity Analysis: The team conducted univariate sensitivity studies on four key meteorological factors: vapor pressure deficit, mean temperature, wind speed, and the saturated water pressure gradient.
  • Scaling: To move from point-based station data to a basin-wide map, they integrated their findings with high-resolution grid data.

Datasets

The study utilized a combination of high-frequency local observations and global reanalysis products:

  • Local Observations (2015): Hourly data from four specialized stations in the heart of the Taklimakan Desert: one NSL site (natural sand) and three CGL sites (west edge, middle, and east edge of a green belt).
  • Validation Data: Evaporating dish data from the National Tazhong Meteorological Station served as the ground-truth "standard" for comparison.
  • ERA5 Reanalysis Data: Hourly data at a 0.25° × 0.25° resolution (including dewpoint, temperature, wind speed, and solar radiation) was used to calculate PET across the broader Tarim Basin.

Key Insights and Ideas

  • The "Cooling" Effect of Greening: Surprisingly, PET is significantly lower in constructed green lands than in natural desert. While plants transpire, the green belts break the wind and increase local humidity, which suppresses the overall "evaporative demand" of the atmosphere more than the bare, wind-swept sand does.
  • Seasonal Sensitivity: The difference between calculated PET and observed evaporation is much larger during the vegetation growing season (April–September). In July, the difference between NSL and CGL PET can reach up to 400 mm.
  • Dust as a Driver: In this region, weather conditions like "floating dust" and "blowing sand" are the primary drivers of PET variability, contributing to 60% and 35% of the weather-related PET fluctuations, respectively.
  • Driver Dynamics: PET in this model is positively influenced by wind speed and vapor pressure deficit (the "dryness" of the air) but showed a negative sensitivity to increases in the 2-m mean temperature and the saturated water pressure gradient within the specific context of the PM-H algorithm's parameters for this region.

Key Results

  • Basin-Wide PET: The mean annual potential evapotranspiration for the Tarim Basin was calculated at 2,750 mm.
  • Model Accuracy: The PM-H model showed high correlation with observed data, with $R^2$ values of 0.952 for natural sand and ~0.94 for green lands.
  • NSL vs. CGL: Monthly mean PET for natural sand was approximately 283.14 mm, while the three green land sites averaged significantly lower at roughly 113 mm.
  • Spatial Distribution: PET is generally higher in the center of the basin and decreases toward the surrounding green oases and mountain fringes.

Limitations and Future Work

  • Temporal Scope: The study focused primarily on data from 2015. The authors noted that equipment damage (specifically irrigation pipe fractures) in 2014 led to missing data, limiting the ability to look at multi-year trends.
  • Spatial Resolution: While 0.25° grid data is a standard for regional studies, it may still be too coarse to capture the micro-climatic nuances of very narrow green belts or patchy desert vegetation.
  • Future Directions: The authors suggest that future research should incorporate more long-term "big data" to analyze how climate change is shifting these PET patterns over decades, particularly as human-led greening efforts expand.

Summary updated

Abstract

As an important part of surface hydrothermal process, the study of evapotranspiration has been paid more attention, and it has an important reference value for the optimal allocation of regional water resources. It hinders the accurate assessment of evapotranspiration in arid zones and its response to climate change because human activities have changed the original underlying surface of deserts and increased the complexity of calculations. In this study, the encrypted observations of natural sand land (hereinafter referred to as NSL) and constructed green land (hereinafter referred to as CGL) in the heartland of the Taklimakan Desert were analyzed by the Penman-Monteith-H (PM-H) numerical model to accurately assess the applicability of the evapotranspiration calculation method to different underlying surfaces, analyze the associated errors, and quantify the effects and contributions of the main impacting meteorological factors on the evapotranspiration calculation. Our study finds that the PM-H algorithm exhibits good applicability in Tarim Basin, where the vegetation growth period significantly affects evapotranspiration. The evapotranspiration calculation is positively affected by the difference between saturated and actual vapor pressures and the 2-m mean wind speed and negatively affected by the 2-m mean temperature and gradient of saturated water pressure. Based on high-resolution grid data, the applicability of the PM-H algorithm has been improved, and our calculations show that the mean annual potential evapotranspiration in the Tarim Basin is 2,750 mm.
BibTeX
@inproceedings{huo2022refined,
  author = {Wen Huo and Xiefei Zhi and Shunqi Hu and Wenyue Cai and Fan Yang and Chenglong Zhou and Ali Mamtimin and Qing He and Honglin Pan and Meiqi Song and Cong Wen and Yu Wang and Xinghua Yang and Lu Meng},
  title = {{Refined assessment of potential evapotranspiration in the tarim basin}},
  booktitle = {Frontiers in Earth Science},
  year = {2022},
  doi = {10.3389/feart.2022.904129},
}
Figure