Comparison of solar-induced chlorophyll fluorescence, light-use efficiency, and process-based GPP models in maize

Accurately quantifying cropland gross primary production (GPP) is of great importance to monitor cropland status and carbon budgets. Satellite-based light-use efficiency (LUE) models and process-based terrestrial biosphere models (TBMs) have been widely used to quantify cropland GPP at different scales in past decades. However, model estimates of GPP are still subject to large uncertainties, especially for croplands. More recently, space-borne solar-induced chlorophyll fluorescence (SIF) has shown the ability to monitor photosynthesis from space, providing new insights into actual photosynthesis monitoring. In this study, we examined the potential of SIF data to describe maize phenology and evaluated three GPP modeling approaches (space-borne SIF retrievals, a LUE-based vegetation photosynthesis model [VPM], and a process-based soil canopy observation of photochemistry and energy flux [SCOPE] model constrained by SIF) at a maize (Zea mays L.) site in Mead, Nebraska, USA. The result shows that SIF captured the seasonal variations (particularly during the early and late growing season) of tower-derived GPP (GPP_EC) much better than did satellite-based vegetation indices (enhanced vegetation index [EVI] and land surface water index [LSWI]). Consequently, SIF was strongly correlated with GPP_EC than were EVI and LSWI. Evaluation of GPP estimates against GPP_EC during the growing season demonstrated that all three modeling approaches provided reasonable estimates of maize GPP, with Pearson's correlation coefficients (r) of 0.97, 0.94, and 0.93 for the SCOPE, VPM, and SIF models, respectively. The SCOPE model provided the best simulation of maize GPP when SIF observations were incorporated through optimizing the key parameter of maximum carboxylation capacity (V-cmax). Our results illustrate the potential of SIF data to offer an additional way to investigate the seasonality of photosynthetic activity, to constrain process-based models for improving GPP estimates, and to reasonably estimate GPP by integrating SIF and GPP_EC data without dependency on climate inputs and satellite-based vegetation indices.