link to homepage

Institute of Bio- and Geosciences

Navigation and service

Algorithms to retrieve land surface temperature in areas of highly dynamic emissivity using thermal infrared data from satellites and drones

Sascha Heinemann

Land surface temperature (LST) is a key variable in order to understand the processes of energetic interactions between the Earth's surface and the atmosphere. This knowledge is fundamental for various environmental research questions, particularly with regard to climate change. The current challenge is to overcome major deviations between retrieved LST data from satellite observations and in situ measurements (e.g. airborne, by drones). An accurate determination of land surface emissivity (LSE) is required to retrieve correct LST products. So far, there are no unsupervised methods to retrieve properly LST and to derive LSE particularly in areas of highly dynamic emissivities.

Therefore, especially for regions with seasonal changes in vegetation cover including various landscape types such as agricultural land, grassland and mixed forests, and for “urban heat islands”, different methods are investigated to identify the most fitting one. LSE is initially derived using NDVI-based threshold methods to distinguish between bare soil, dense vegetation cover and pixel composed of bare soil and vegetation. Subsequently, LST is calculated with Planck’s radiation law.

 Thermal infrared image of Campus Klein-Altendorf (2017-06-23)