Enhancing carbon flux estimation in a crop growth model by integrating UAS-derived leaf area index

Accurate estimation of agroecosystem carbon fluxes is essential for assessing cropland sustainability and climate resilience. This study integrates Leaf Area Index (LAI) retrieval from Radiative Transfer Model (RTM) inversion into AgroC, an agroecosystem model, from Unmanned Aerial System (UAS) platform to enhance carbon fluxes estimates, including Gross Primary Production (GPP), Net Ecosystem Exchange (NEE), and Total Ecosystem Respiration (TER). By replacing the internally developed LAI in the AgroC model with interpolated LAI time series derived from UAS, improved spatiotemporal representativeness of agroecosystem carbon fluxes is observed under both the Farquhar-von Caemmerer-Berry (FvCB) and the Light Use Efficiency (LUE) photosynthesis approaches. Temporally, the highest GPP accuracy was achieved by the AgroC_FvCB model integrated with UAS-derived LAI (RMSE = 3.19 gC m⁻² d⁻¹, KGE = 0.89), while the best NEE estimation was obtained with the AgroC_LUE model integrated with UAS-derived LAI (RMSE = 2.10 gC m⁻² d⁻¹, KGE = 0.89). Spatially, the superior performance of the AgroC_FvCB model in integrating UAS-derived LAI enabled high-resolution (1 m) mapping of GPP and NEE, effectively capturing within-field spatial variations in a winter wheat field. The daily Pearson correlation coefficient $\left(r\right)$ overtime ranged from 0.16 in non-vegetated areas to 0.94 in vegetated zones for GPP, and up to 0.88 for NEE. Despite the advantages taking physical basis in RTM inversion for LAI retrieval and biochemical constraints considered in FvCB approach, the limitation in TER improvement requires further investigation to refine RTM-AgroC coupling for cropland carbon fluxes modelling using UAS platforms.