Improving process-based modeling of crop production in the DayCent ecosystem model with solar-induced chlorophyll fluorescence

• New ecosystem model integrated with a mechanistic light response framework was evaluated. • Coupled approach accounting for seasonal Φ PSIImax is more appropriate for simulating GPP . • Improved model simulated GPP more accurately than conventional RUE approaches. Solar-induced chlorophyll fluorescence (SIF), as the red and far‐red light emitted from excited chlorophyll‐a molecules during photosynthesis, has been increasingly used for estimating gross primary production (GPP). DayCent is a process-based ecosystem model that mechanistically simulates time-dependent biogeochemical processes of the plant and soil system at a daily time-step. The present study explored a series of key modifications to develop photosynthetic equations with the Mechanistic Light Response (MLR) framework. We used published ground‐based measurements derived from five cropland sites with multiple years of data in the central region of United States to calibrate and evaluate the MLR framework in DayCent. Three versions of the DayCent SIF model were developed and tested based on variation in algorithms and main parameters in MLR equation. The best fit model version for predicting seasonal GPP rates used temperature to capture the seasonal variation of the maximum photochemical efficiency of photosystem II (Φ PSIImax ) and directly calculated the intercellular CO 2 concentration (C i ) based on Eco-Evolutionary Optimality (EEO) theory, with index of agreement (IA) varying between 0.76 and 0.88 and root mean square error (RMSE) between 3.93 and 6.76 g C m −2 d -1 . This version of the model also has significantly better agreement with measured GPP, as demonstrated by higher IA value and lower RMSE value when compared to conventional Radiation Use Efficiency (RUE) approaches, even when informed by the MODIS Enhanced Vegetation Index (EVI). This study demonstrates that a mechanistic modeling framework informed by SIF observations can improve process-based modeling of crop production.