How to obtain a highly accurate dataset of the snow surface temperature with a thermal infrared camera?

Snow plays a critical role in alpine areas, influencing the local climate and serving as a crucial water reservoir for downstream ecosystems and human activities. The surface temperature of snow provides many insights about the current state of the snowpack and helps water storage estimations. While satellites are regularly used to measure surface temperature of snow over alpine areas, accurate measurements are still difficult to retrieve from space, and calibration-validation initiatives over snow-covered areas are scarce. In this context, we produced a two-winter timeseries of approximately 130,000 maps of the radiative surface temperature of snow acquired with an uncooled Thermal Infrared camera. TIR images were acquired November 2021 to May 2022 and February to May 2023 at the Col du Lautaret, 2057 m a.sl. in the French Alps. During the first season, the camera operated in the off-the-shelf configuration, with a rough thermal regulation (7°C - 39°C) resulted in timeseries of snow surface temperature maps with an absolute accuracy <1.25 K. The large variations of the camera’s internal temperature were identified as the main source of error. An improved setup using a thermoelectric cooler to stabilize the internal temperature was therefore developed for the second campaign, while comprehensive laboratory experiments led to a thorough characterization of the physical properties of the TIR camera and its calibration. A meticulous processing includes radiometric processing, orthorectification and a filter for foggy and snowy images. The validation against precision TIR radiometers deployed in the camera’s field of view results in an estimated absolute accuracy <0.7 K for spring 2023. The efforts to stabilize the internal temperature of the TIR camera therefore led to a notable improvement of the accuracy. This methodology represents a significant advance in the capacity to map the snow surface temperature over complex terrain, overcoming the issues found to get accurate thermal infrared images of absolute temperature discussed in previous studies. The methodology, as well as the resulting timeseries, will be useful for the investigation of the surface energy budget of snow and for the calibration/validation of satellite thermal infrared products such as Landsat, ECOSTRESS and, starting in 2025, TRISHNA over snow.