3SEB: a remote sensing-based three-source energy balance model to improve global estimations of evapotranspiration in semi-arid tree-grass ecosystems

Authors: BURCHARD-LEVINE, V. - Spanish National Research Council; NIETO, H. - University Of Alcala; RIANO, D. - Spanish National Research Council; Kustas, William - Bill; MIGLIAVACCA, M. - Max Planck Institute For Biogeochemistry; EL-MADANY, T. - Max Planck Institute For Biogeochemistry; NELSON, J. - Ifapa Centro Alameda Del Obispo; ANDREU, A. - Ifapa Centro Alameda Del Obispo; CARRARA, A. - Collaborator; BERINGER, J. - Monash University; BALDOCCHI, D. - University Of California; PILAR MARTIN, M. - Spanish National Research Council

Submitted to: Global Change Biology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/1/2021
Publication Date: 2/1/2022
Citation: Burchard-Levine, V., Nieto, H., Riano, D., Kustas, W.P., Migliavacca, M., El-Madany, T., Nelson, J., Andreu, A., Carrara, A., Beringer, J., Baldocchi, D., Pilar Martin, M. 2022. 3SEB: a remote sensing-based three-source energy balance model to improve global estimations of evapotranspiration in semi-arid tree-grass ecosystems. Global Change Biology. https://doi.org/https://doi.org/10.1111/gcb.16002.
DOI: https://doi.org/10.1111/gcb.16002

Interpretive Summary: Water resources in semi-arid and savanna ecosystems, representing nearly 15% of the global land surface, are projected to be disproportionately affected by climate and land use change. Remote sensing models of water use or evapotranspiration (ET) have been developed to monitor changes in water resources; however, these models do not sufficiently represent the more complex features of heterogenous semi-arid and savanna landscapes. This study developed and evaluated the thermal-based three-source energy balance (3SEB) model, accommodating an additional vegetation source within the well-known two-source energy balance (TSEB) model. The 3SEB model was evaluated over four semi-arid sites in Australia, Spain and the U.S. with variability in canopy cover and rainfall magnitudes. The proposed model was shown to be robust by accurately simulating ET and related energy fluxes in all four sites and outperformed the TSEB model by explicitly partitioning ET between the overstory and understory vegetation and soil sources. These promising results indicate the potential for the 3SEB model to improve ET and energy flux estimation over complex semi-arid ecosystems, and enhance our capabilities in monitoring drought and other environmental stressors on these semi-arid ecosystems.

Technical Abstract: The uncertainty of remote sensing-based evapotranspiration (ET) retrievals over savannas, or tree-grass ecosystems (TGEs), is a well documented issue. The complex structural and phenological features of these water-limited landscapes provide different challenges compared to more homogeneous and energy-limited ecosystems. The dual vegetation strata, the grass dominated understory and tree dominated overstory, have distinct structural, physiological and phenological characteristics that should be considered within modeling schemes. This study proposes the use of a thermal-based three source energy balance (3SEB) model, accommodating an additional vegetation source within the well-known two-source energy balance (TSEB) model (Norman et al. 1995). 3SEB was evaluated over four TGE sites in Australia, Spain and USA, with variability in canopy cover and rainfall (P) magnitudes. The proposed model was shown to be robust by accurately simulating ET and related energy fluxes in all four sites (ET RMSD: ~0.4 mm/day) and improving over both TSEB and a seasonally changing TSEB (TSEB-2S). In addition, 3SEB inherently partitions water and energy fluxes between the overstory, understory and soil sources. The modelled ET partitioning (T/ET) correlated well with eddy-covariance (EC) based transpiration (T) estimates (r > 0.76). The total ecosystem T/ET partitioning was also shown to be positively related to both P and leaf area index (LAI), however flux patterns from each vegetation source (tree overstory and grass understory) had contrasting relations with respect to monthly P. This demonstrated the importance in decomposing total ET into the different vegetation sources, as they have distinct survival strategies, and hence, different relations to seasonal water availability. These promising results indicate the potential for 3SEB to improve ET and energy flux estimations over complex TGEs, representing nearly 15% of the global surface. This may contribute to enhance our understanding of these landscapes to improve drought monitoring and their responses to climate change feedbacks