Separating near surface thermal inertia signals from a thermal time series by standardized principal component analysis

Abstract

Principal component analysis has been applied to remote sensing data to identify spatiotemporal patterns in a time series of images. Thermal inertia is a surface property that relates well to shallow surface thermal and physical properties. Mapping thermal inertia requires quantifying surface energy balance components and soil heat flux, both of which are difficult to measure remotely. This article describes a method to map soil thermal inertia using principal component analysis applied to a time series of thermal infrared images and it also assesses how sensitive this method is to the time intervals between images. Standardized principal component analysis (SPCA) was applied to thermal infrared images captured at half-hour intervals during a complete diurnal cycle. Shallow surface thermal properties accounted for 45%, 82% and 66% of the spatiotemporal variation in surface temperature observed during the heating phase, cooling phase and over the total diurnal cycle respectively. The remaining 55%, 18% and 34% of the variation was attributed to transient effects such as shadows, surface roughness and background noise. Signals related to thermal inertia explained 18% of total variation observed in a complete diurnal cycle and 7% of variation in the cooling series. The SPCA method was found useful to separate critical information such as timing and amplitude of maximum surface temperature variation from delays related to differential heating induced by micro-topography. For the field conditions experienced in this study, decreased temporal resolution when sampling intervals were greater than an hour significantly reduced the quality of results.