Thermal spectrum solar energy is absorbed by soil during daylight and is emitted during night. Time domain variability of soil surface temperature is a function of subsurface heat capacity and conductivity conditions. Soils that reach their maximum emission at the same time are likely to have the same subsurface arrangements. Low temporal resolution remote sensing hinders constructing soil thermal profiles. The aim of this research is the visualization of three dimensional (3D) patterns in time and space of soil surface temperature development based on high temporal resolution thermal time series, which reflects soil subsurface vertical variability and influences soil delineation (tomography).

The surveyed field is located in southwest Ontario; in an area dominated by artificially drained gleyed soils developed on fine texture lacustrine deposits. Seven images were captured at approximately one-hour time interval using a thermal video camera mounted on a blimp at height of 300 feet during soil thermal emission by nighttime. These images were registered on to a base image using a polynomial model. A spatio-temporal data cube was generated using a cubic convolve of 5x5x5 pixel dimensions, with equal weighting applied to the 124 surrounding pixels.

A fuzzy k means classification is used to classify the time series to several classes grouped into: vegetation canopy, normal and compacted soil. The 3D distributions of maximum emission for each identified class were reconstructed in the spatio-temporal cube by applying a threshold based 3D surface visualization. The distribution of maximum emission of each class is used as an exclusive threshold with the boundary determined at various time intervals. Time domain boundaries were used to segregate soils in to different strata. The reliability of stratification is being tested using ancillary data and field observation.