Predicting land-surface specific humidity from radiative temperature and ambient weather for evapotranspiration modelling: Lessons from South Australian field sites

Land-surface specific humidity is crucial for estimating evapotranspiration (ET) using the Maximum Entropy Production (MEP) method. However, acquiring relevant data, particularly the spatially varying land-surface specific humidity, can be challenging. Here, we show that the deviation of land-surface specific humidity from the ambient specific humidity can be estimated using surface radiative temperature and ambient micrometeorological variables (referred to as the Tr-Weather method). We tested this method at five sites in South Australia with varying vegetation and topography. The results indicate that the Tr-Weather method generally performs the best for early afternoon. The performance varies with seasons, with better results for summer and autumn. Slope and aspect change the timing of optimal predictions, particularly in areas with significant topographic variations. Additionally, this method effectively predicts spatial distribution of the land-surface specific humidity by integrating drone-derived temperature and ambient meteorological data, with an R² value of 0.96. For MEP-based understory ET modelling, the Tr-Weather method outperforms the substituted specific humidity from nearby weather stations, especially under sunny conditions where the MEP ET model using ambient specific humidity tends to underestimate ET. The method is empirical and was developed based on observations in two different environments, further research is required to extend and validate the Tr-Weather approach over other bioclimate zones. Nevertheless, our findings demonstrate the potential of applying the Tr-Weather method, supported by drones and high-resolution satellite data, to advance MEP-based ET modelling across broader landscapes.