Lake surface water temperature (LSWT), which varies spatially and temporarily, reflects meteorological and climatological forcing more than any other physical lake parameter. There are different data sources for LSWT mapping, including remote sensing and in situ measurements. Depending on cloud cover, satellite data can depict large-scale thermal patterns, but not the meso- or small scale processes. Thermography at the meso-scale provides the ability to resolve (and hence ground-truth) satellite imagery at the sub-pixel scale. A Balloon Launched Imaging and Monitoring Platform (BLIMP) was used to measure the LSWT at the meso-scale. The BLIMP consists of a small balloon tethered to a boat and which is equipped with thermal and RGB cameras, as well as other instrumentation for location and communication. Simultaneous ground-truthing of the BLIMP data was achieved using an autonomous craft measuring a variety of data, including in situ surface/near surface temperatures, radiation and meteorological data. The latent and sensible surface heat fluxes were calculated using the bulk parameterization algorithm based on similarity theory. A feature matching-based algorithm was implemented to create composite thermal images. Results are presented for the daily stratified low wind speed (up to 3 m/s) conditions over Lake Geneva. Two field campaigns, each of ~6 h from 18 March and 19 July 2016 are presented. The mesoscale temperature field (~1-m pixel resolution) had a range and standard deviation of 2.4°C and 0.3°C, respectively, over a 1-km2 area (typical satellite pixel size). Interestingly, at the sub-pixel scale, various temporal and spatial thermal eddy structures are evident – an obvious example being streaks in the along-wind direction during March, which we hypothesize are caused by the steady ~3 h wind condition. The results also show that the spatial variability of the estimated total heat flux is due to the corresponding variability of the longwave cooling from the water surface and the latent heat flux.