Remotely Sensed Soil Temperatures Beneath Snow-Free Skin-Surface Using Thermal Observations from Tandem Polar-Orbiting Satellites: an Analytical Three-Time-scale Model

Wenfeng Zhan,Ji Zhou,Weimin Ju,Manchun Li,Inge Sandholt,James Voogt,Chao Yu
DOI: https://doi.org/10.1016/j.rse.2013.12.004
IF: 13.5
2014-01-01
Remote Sensing of Environment
Abstract:Subsurface soil temperature is a key variable of land surface processes and not only responds to but also modulates the interactions of energy fluxes at the Earth's surface. Thermal remote sensing has traditionally been regarded as incapable of detecting the soil temperature beneath the skin-surface. This study shows that thermal remote sensing can be used to estimate soil temperatures. Our results provide insights into thermal observations collected with tandem polar-orbiting satellites when used toward obtaining soil temperatures under clear-sky conditions without the use of any ground-based information or field-measured soil properties.We designed an analytical three-time-scale (3-scale, for short) model, dividing the annual cycle of soil temperatures into three subcycles: the annual temperature cycle (ATC), which represents the daily-averaged temperature; the diurnal temperature cycle (DTC), which represents the instantaneous temperature; and the weather-change temperature cycle (WTC), which is divided into two parts to represent both the daily-averaged (WTCavg) and the instantaneous temperature (WTCinst). The DTC and WTCinst were further parameterized into four undetermined variables, including the daily-averaged temperature, thermal inertia, upward surface flux factor, and day-to-day change rate. Thus, under clear-sky conditions, the four thermal measurements in a diurnal cycle recorded with tandem polar-orbiting satellites are sufficient for reconstructing the DTC of both land surface and soil temperatures. Polar-orbiting satellite data from MODIS are used to show the model's capability. The results demonstrate that soil temperatures with a spatial resolution of 1km under snow-free conditions can be generated at any time of a clear-sky day. Validation is performed by using a comparison between the MODIS-inverted and ground-based soil temperatures. The comparison shows that the accuracy of inverted soil temperatures lies between 0.3 and 2.5K with an average of approximately 1.5K. These results open a new frontier in the application of thermal remote sensing wherein soil temperatures with high spatial and temporal resolutions can be remotely estimated.
What problem does this paper attempt to address?