Differences in diffuse photosynthetically active radiation effects on cropland light use efficiency calculated via contemporary remote sensing and crop production models

Diffuse photosynthetically active radiation (PAR_diff) is instrumental to the light use efficiency (LUE) of vegetation. Accurately assessing the impact of PAR_diff on crop LUE can better our understanding of the carbon cycle in cropland ecosystems. LUE estimates from six remote sensing models (including four big-leaf models and two two-leaf models) and two crop production models were compared with measured FLUXNET LUE data from cropland sites under different PAR_diff fraction (FDIFF_PAR) intervals. Compared with the FLUXNET observations, the Eddy Covariance-Light Use Efficiency (EC-LUEa) model exhibited the best LUE estimation (R² = 0.250, RMSE = 0.868 gC·MJ⁻¹, and Bias = −0.005 gC·MJ⁻¹) owing to the use of more accurate calculation scheme of environmental stress factors. LUEs calculated from FLUXNET observational data were positively correlated with FDIFF_PAR, but only LUEs simulated by the Moderate Resolution Imaging Spectroradiometer Photosynthesis (MOD17) and Two-Leaf Light Use Efficiency (TL-LUE) models increased with increasing FDIFF_PAR. This is attributed to the fact that the MOD17 model divides the crop growth types into cereal and broadleaf, while the TL-LUE model considers the change of light interception with increased FDIFF_PAR. Furthermore, the maximum LUE (LUE_max) increased with FDIFF_PAR at FLUXNET observational sites, but the eight models could not capture the effects of PAR_diff on the crop LUE_max. Among the eight models, the LUE_max–FDIFF_PAR relationship simulated by the two-leaf models fluctuated because the crops were divided into sunlit and shaded leaves, while the big-leaf and crop production models used a constant LUE_max and showed a constant LUE_max–FDIFF_PAR relationship. Additionally, big-leaf models performed better than two-leaf models for gross primary production (GPP) simulation in the cropland ecosystem, which is related to the planting density and vegetation structure. These results demonstrate the importance of considering the impact of FDIFF_PAR on LUE_max in LUE modeling.