Land Surface Temperature (LST) is one of the most important variables giving access to water and energy budgets in the continuum soil-biosphere-atmosphere. Many applications requires high spatial and temporal resolution LST data. The development of remote sensing instrument make possible to retrieve thermal data from satellite sensors at different temporal and spatial resolutions. However, satellite thermal data are either high spatial resolution or high temporal one. To assess the utility of LST satellite data in monitoring hydric and energy budgets at appropriate scales, solutions to combine these multi-scale and multi-temporal data have been so far proposed. This paper presents a new methodology for CSR thermal data sharpening based on Genetic Particle Smoother GPS. The GPS Thermal Sharpener (GPSTS) was implemented and applied to a land surface dynamic model able to simulate prior end-member temperatures. The micro-meteorological data rely on the French “Crau-Camargue” region landscape and climate for the year 2006. GPSTS performances were evaluated against prior model estimations and temporal sampling. Results show that the GPSTS outperforms the prior model estimations. Nonetheless, these performances are very sensitive to observation error statistics and observation period and frequency.