Accurately quantifying gross primary production (GPP) is crucial for monitoring carbon cycles, assessing the global carbon budget, and predicting vegetation stress in a changing climate. Solar-induced chlorophyll fluorescence (SIF), an optical signal emitted by chlorophyll a during photosynthesis under solar illumination, serves as a remote sensing proxy for mechanistically understanding photosynthesis from space. Realistic model parameterization of SIF is critical for constraining GPP estimations. However, accurate SIF parameterizations and simulations depend on model structure (mathematical formulation of processes) and parameters (coefficients in these mathematical formulations). The degree to which model structure and parameters impact SIF simulations has never been evaluated with independent measurements. In this thesis, I utilized the Soil Canopy Observation Photosynthesis Energy (SCOPE) radiative transfer model to simulate SIF and GPP at three forest sites within the National Ecological Observatory Network (NEON). First, I evaluated the impact of model structure on SIF simulations, through two sets of simulations with different SIF parameterizations: 1) based on an empirical non-photochemical quenching (NPQ) parameterization, and 2) utilizing a mechanistic light reaction (MLR) model by modeling the fraction of open Photosystem II (PSII) reaction centers (qL). Second, to examine the impact of model parameters on SIF (and GPP) simulations, I compared simulations using parameters that are 1) plant functional type (PFT)-specific (denoted as ctrl), and 2) foliar trait-specific measurements (denoted as para). Further, I investigated the impact of subgrid spatial heterogeneity of coarse-resolution SIF observations and the resulting scale mismatch with model simulations. To achieve this I compared simulations conducted at tower footprint with a single land cover (denoted as baseline) vs land cover (LC)-specific simulations within a 0.05° satellite grid. My analyses revealed that qL-based SIF parameterization outperformed NPQ-based parameterization in capturing the 16-day SIF time series from Orbiting Carbon Observatory-2 (OCO-2) and TROPOspheric Monitoring Instrument (TROPOMI) satellite observations. In addition, foliar trait-specific parameters significantly improved both the simulated GPP and SIF (when using qL-based scheme). In particular, the choice of SIF model structure impacted the sensitivity of simulated SIF to fundamental photosynthetic parameters; for example, the qL-based SIF model has a higher sensitivity to variations in the maximum carboxylation rate (Vcmax). Finally, accounting for sub-grid spatial heterogeneity, i.e., explicitly prescribing LAI and parameters (foliar trait-based) separately for each end member land cover, significantly improved agreement between the model simulations and satellite observations. In summary, this thesis highlighted the importance of physiologically meaningful combinations of model structure and foliar trait-specific parameters, as well as the consideration of sub-grid spatial heterogeneity in enhancing the realism and accuracy of SIF simulations. These insights are vital for advancing our knowledge in estimating, understanding, and predicting SIF and GPP, and ultimately the global carbon budget in response to climate change.