DATA FUSION, COMPRESSION, AND VISUALIZATION OF THERMAL AND VISIBLE IMAGERY FOR REMOTE ANALYSIS OF GEOLOGIC SURFACES ON EARTH AND MARS

Global datasets of high-resolution thermal infrared (TIR) and visible to near-infrared (VNIR) satellite imagery provide opportunities for mapping planetary surface properties at resolutions that can be directly applied to field observations, with coverage that allows for analysis at regional to global scales. This information can be ideal for investigating geologic processes, climate variables, and surface history on the terrestrial planets, although advanced processing and analytical tools are needed to calibrate and display the data in a way that is useful for field application. To facilitate the use of these multispectral, multitemporal data in a wider scientific community, a new data-fusion technique developed using Mars Thermal Emission Imaging System (THEMIS) multispectral imagery and modified for use with the Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) provides a data product that can be manipulated using commonly available software to map and interpret surface physical properties such as sediment grain size, bedrock exposure, and water content of surface materials.

The term thermal morphology is used here to describe the combination of daytime visible reflectance with nighttime brightness temperature, in which the physical properties controlling the diurnal temperature and albedo can be directly interpreted. Daytime visible images produce scenes in which the reflectivity, topography, and surface roughness provide the majority of variation within a field of view (Figure 1a). Daytime thermal images are similar to daytime visible, in which the albedo and morphology dominate the temperature variation within a scene (Figure 1b). Nighttime thermal images display information related primarily to the thermal inertia of materials, in which albedo and topographic information is significantly subdued (Figure 1c). Thermal inertia represents the ability of near-surface materials to absorb solar energy during the day, conduct it into the sub-surface, and then release that energy throughout the night. A combination of these datasets results in a striking image where colorized nighttime thermal information is draped over daytime data (Figure 1d).

For application to ASTER, this involves the compression of co-registered and calibrated 3-band VNIR and 5-band TIR radiance into a single, RGB-color, byte image (geotiff). In the thermal infrared emissivity/temperature separation, 5-band nighttime observations are combined to produce a one-band real number image. Given the normal range of diurnally-varying temperatures of natural surfaces on Earth, the observed range can be linearly converted to byte data range (0-255), and retain 0.1º C temperature resolution. Daytime observations are equally compressed in data volume to present the most thermophysically-significant information in a compressed format. ASTER 3-band VNIR data is integrated and converted to top of atmosphere calibrated reflectance, producing a single image of minimum data resolution in albedo of 0.004. Conversion of nighttime temperature to a standardized hue-based color gradient allows temperature to be draped over grayscale visible reflectivity resulting in a color image which displays the information from those two datasets. Retrieval of temperate and albedo values can be made visually, with the aid of a color gradient and greyscale. Digital separation of temperature and albedo can be performed by converting the RGB to HSI (hue, saturation, intensity) format, in which hue is converted to temperature information and intensity to albedo. This method allows eight bands of data to be compressed into a 3-band byte image, and displayed in a format that can be interpreted in the field while retaining quantitative information for detailed analysis.

Figure1. ASTER imagery from the Steen Mountains in southeastern Oregon, providing four perspectives used in thermophysical analysis. Visible albedo (a) is controlled by the reflectivity, roughness, and topography of the surface. Daytime brightness temperature (b) provides an image similar to VNIR reflectance, since it is primarily controlled by the albedo, and insolation. Nighttime brightness temperature (c) is primarily a function of the thermal inertia, which can be used to constrain the effective sediment grain size, material conductivity, and surface water content. Thermalmorphology (d) is the combination of greyscale visible (a) and colorized nighttime temperature (c) images.