The thermal regime of lakes and reservoirs is a major driver of their functioning. However, water temperature measurements are scarce and time-series are frequently interrupted by missing data in many lakes and reservoirs. The aim of this paper is to assess whether satellite imagery combined with three-dimensional (3D) modelling can overcome this limitation. We assessed the performance of a 3D model (Delft3D-Flow) and compared simulated temperature to satellite data on a reservoir in a semi-arid region, Karaoun reservoir, Lebanon. Surface temperatures were retrieved from Landsat 8, atmospherically corrected using a single channel algorithm and adjusted with in situ measurements as a proxy of bulk temperatures. With very limited calibration, the model reproduced water level fluctuations, water temperature, stratification and mixing, with low discrepancies from measurements. Satellite temperatures were found to be in good agreement with observations and simulations. Satellite temperature distributions across the reservoir exhibited low spatial heterogeneity. The 3D model partly reproduced this spatial distribution of surface temperature but generally showed good ability to simulate the thermal regime of the reservoir with a limited data set for initial conditions and for hydrological and meteorological forcing, and with limited calibration. For validating 3D hydrodynamic models, satellite temperatures constitute a valuable source of data which are complementary to point measurements. Coupling satellite imagery and modelling can improve operational surveys of poorly monitored lakes and reservoirs, and enhance knowledge of their thermal functioning.