Gou et al. (2025) Predicting land-surface specific humidity from radiative temperature and ambient weather for evapotranspiration modelling: Lessons from South Australian field sites

Identification

• Journal: Agricultural and Forest Meteorology
• Year: 2025
• Date: 2025-10-10
• Authors: Jianfeng Gou, Wenjie Liu, Jessica R. Thompson, Okke Batelaan, Hailong Wang, Karina Gutierrez, Juliette Woods, Huade Guan
• DOI: 10.1016/j.agrformet.2025.110878

Research Groups

• National Centre for Groundwater Research and Training, College of Science and Engineering, Flinders University, Adelaide, Australia
• Faculty of Architecture and Civil Engineering, Huaiyin Institute of Technology, Huai’an, China
• College of Hydrology and Water Resources, Hohai University, Nanjing, China
• Murray-Darling Basin Authority, Mildura, Australia
• School of Civil Engineering, Sun Yat-sen University, Guangzhou, China
• Department for Environment and Water, South Australia Government, Adelaide, Australia

Short Summary

This study develops an empirical "Tr-Weather" method to estimate land-surface specific humidity using surface radiative temperature and ambient micrometeorological variables, demonstrating its superior performance for evapotranspiration (ET) modeling, especially under sunny conditions, compared to using ambient specific humidity.

Objective

• To develop an empirical method (Tr-Weather) integrating surface radiative temperature and ambient weather conditions for estimating land-surface specific humidity.
• To investigate how local surface specific humidity deviation (SSHD) from ambient conditions depends on radiative temperature and ambient micrometeorological variables.
• To determine the optimal time of day for this dependency and how it varies with seasons, vegetation cover, and topography.
• To assess if spatially distributed radiative temperature (e.g., drone-derived) and ambient meteorological data can effectively predict the spatial variability of land-surface specific humidity.
• To evaluate the contribution of Tr-Weather estimated land-surface specific humidity to Maximum Entropy Production (MEP) ET simulation compared to using ambient specific humidity from nearby weather stations.

Study Configuration

• Spatial Scale: Five field sites in South Australia (three in Bookpurnong, two in Mount Wilson) with varying vegetation and topography. Drone imagery for the HF site with 1 meter (m) spatial resolution.
• Temporal Scale: Bookpurnong sites: October 2021 to December 2022. Mount Wilson sites: June 2013 to July 2014. Ambient micrometeorological data at 30-minute intervals, surface radiative temperature at 15-minute intervals. Drone surveys on March 4, April 10, and May 7, 2022.

Methodology and Data

• Models used:
  • Tr-Weather method: Empirical method for estimating land-surface specific humidity deviation (SSHD) using stepwise linear regression (SLR).
  • Maximum Entropy Production (MEP) method: Used for evapotranspiration (ET) modeling.
  • Double-Shading Transposition (DST) model: Applied to calculate under-canopy downwelling shortwave radiation.
• Data sources:
  • Ambient micrometeorological data: Downwelling shortwave radiation, ambient temperature, relative humidity (3 m or 12 m above ground), daily and cumulative precipitation, air pressure from field sites and nearby Bureau of Meteorology stations.
  • Surface radiative temperature: Measured by Apogee SI-111SS Infrared Radiometers (1.5 m above ground) and DJI Matrice 300 RTK drone with Zenmuse H20T thermal camera.
  • Surface relative humidity and temperature: Measured by iButton sensors (within 10-30 cm from the ground).
  • Ancillary data: LiDAR-derived canopy height and structure data for drone temperature correction.

Main Results

• The Tr-Weather method generally performs best for early afternoon (12:00–15:00), with an average R² value of 0.67 at 12:00.
• Method performance varies seasonally, yielding better predictions in summer and autumn than in winter and spring.
• Slope and aspect influence the timing of optimal predictions, particularly in areas with significant topographic variations.
• The method effectively predicts the spatial distribution of land-surface specific humidity by integrating drone-derived temperature and ambient meteorological data, achieving an R² value of 0.96.
• For MEP-based understory ET modeling, the Tr-Weather method significantly outperforms the substituted specific humidity from nearby weather stations (ambient qa method), especially under sunny conditions where the qa method tends to underestimate ET.
  • Tr-Weather method: R² = 0.89, regression slope = 0.96.
  • Ambient qa method: R² = 0.84, regression slope = 0.69.
• Ambient temperature and under-canopy downwelling shortwave radiation are the most influential variables, jointly explaining over 50% of SSHD variance. Radiative temperature contributes 14.5% in summer/autumn and 9.6% in winter/spring.

Contributions

• Development of the empirical Tr-Weather method for estimating land-surface specific humidity from readily available radiative temperature and ambient weather data.
• Identification of optimal daytime periods (early afternoon) for predicting land-surface specific humidity, providing guidance for drone-based remote sensing applications.
• Significant improvement in the accuracy of MEP-based evapotranspiration (ET) modeling, particularly for understory ET and under sunny conditions, by providing more accurate land-surface specific humidity inputs.
• Demonstration of the Tr-Weather method's spatial generalizability through integration with drone-derived thermal imagery, enabling high-resolution mapping of land-surface specific humidity.
• Insights into the seasonal and topographic variations of land-surface specific humidity dependency on meteorological variables.

Funding

• Murray-Darling Basin Authority
• National Centre for Groundwater Research and Training, Australia
• China Scholarship Council

Citation

@article{Gou2025Predicting,
  author = {Gou, Jianfeng and Liu, Wenjie and Thompson, Jessica R. and Batelaan, Okke and Wang, Hailong and Gutierrez, Karina and Woods, Juliette and Guan, Huade},
  title = {Predicting land-surface specific humidity from radiative temperature and ambient weather for evapotranspiration modelling: Lessons from South Australian field sites},
  journal = {Agricultural and Forest Meteorology},
  year = {2025},
  doi = {10.1016/j.agrformet.2025.110878},
  url = {https://doi.org/10.1016/j.agrformet.2025.110878}
}