Abstract. Far-red Sun-Induced chlorophyll Fluorescence (SIF) is increasingly used as a proxy of vegetation Gross Primary Production (GPP) across different ecosystems and at spatiotemporal resolutions going from proximal to satellite-based remote sensing measurements. However, the use of SIF to probe variations in GPP in forests is challenged by (1) confounding factors such as canopy structure and sun-canopy geometry, and by (2) leaf physiological and biochemical properties along with abiotic factors (light intensity, temperature, soil water content, atmospheric vapour pressure deficit, etc.) that can influence SIF and GPP in a different way. To provide insights into understanding the complex drivers of GPP and SIF variations and of their relationships, we examined how SIF and GPP changed at daily and seasonal scales and how canopy structure and environmental conditions affected SIF and GPP relationships in a deciduous oak forest. To do so, we combined canopy scale SIF measurements, spectral vegetation indices, environmental variables measurements, including diffuse and direct radiations in the spectral range of the Photosynthetically Active Radiation (PAR), air and canopy temperature, soil water content (SWC), atmospheric Vapour Pressure Deficit (VPD), and GPP estimated from eddy covariance measurements. Canopy chlorophyll fluorescence was also measured using an active system with an artificial light source, referred to as LIF (LED Induced chlorophyll Fluorescence) hereafter. Further, Random Forest (RF) models were used to predict SIF and GPP and to analyse the responses of SIF and GPP to environmental drivers. The results show that both SIF and GPP variations and their relationships were dependent on the temporal scale considered. At the seasonal scale, The data show that leaf and canopy properties variations, seasonal cycle of PAR, and other abiotic factors such as VPD and SWC control not only SIF and GPP variations, but also their relationships. Further, during extreme weather conditions (heatwaves observed in 2022 in: mid-June (DOY: 166-169), mid-July (DOY: 196-199), and early August (DOY: 218-224)), we observed that SIF and reflectance-based Vegetation Indices (VIs), such as Normalized Difference Vegetation Index (NDVI) and Near-Infrared Reflectance of vegetation index (NIRv), and also SIF and PAR are uncorrelated, while GPP, SIF, passive SIF yield (SIFy) and active chlorophyll fluorescence yield (FyieldLIF) strongly decreased. This indicates that during these severe abiotic conditions SIF stayed a usable proxy of GPP, while VIs cannot be used to track changes in vegetation physiology. This specific response of SIF compared to VIs underlined the interest of SIF to monitor GPP under severe abiotic conditions. At the diurnal timescale, the results also revealed that the saturation of the relationship between GPP and SIF was not only dependent on PAR, but also on the fraction of diffuse to total PAR, as well as on VPD, SWC, and air and canopy temperature. The other key finding was that sun geometry angles had strong effects on GPP and SIF variations. This result highlights that using ground-based SIF measurements to validate satellite measurements at coarse spatial and temporal resolutions can therefore be very difficult, due to confounding factors whose effects are significant and may vary from one site to another, especially in forest canopies.