Assessing Lava Flow Subpixel Surface Roughness and Particle Size Distribution for Improved Thermal Inertia Interpretations

Abstract

Apparent thermal inertia (ATI) is a remote sensing-based thermophysical approximation used to estimate surface properties such as particle size or soil moisture, but is subject to oversimplifications if uniform surface materials are assumed. Geological surfaces realistically contain multiple types of materials that are represented by mixed pixels in a digital image that can dramatically affect the derived thermal response. Thus, the current surface uniformity assumption can lead to erroneous calculations. To define how these mixed particle size surfaces affect ATI, a multi-instrument, multi-spectral study was conducted at the North Coulee rhyolite flow, Mono Domes (California). This flow is compositionally homogenous with particle sizes ranging from silt size to boulders, making it an ideal location to understand the relationship between particle size distributions and ATI. Multispectral data from orbital sensors with increasing spatial resolutions were analyzed in combination with samples, GPS, and photogrammetry data collected in the field. The surface particle size characteristics divided into three broad categories (fine, moderate, and coarse) were mapped based on WorldView-2 visible data and field observations. Broad categories were validated using a 3-dimensional point cloud derived from structure-from-motion (SfM) methods using field photographs. The areal percentage of each category within an ATI pixel was derived to populate a lookup table (LUT) with the corresponding ATI values to quantify the effect of mixed pixels. Lower ATI values are dominated by fine particles (sand and dust) as expected, however, surfaces with the highest values were predominately moderate-sized cobbles. Pixels with a high areal percentage of coarse sizes display an intermediate ATI value, suggesting that either self-shadowing or trapping of fines lowers the ATI value. Alternatively, areas with a majority of moderate particle sizes have less shadowing and efficient vertical sorting of smaller material, as seen in field data, leaving a thermal derived response indicative of the dominating particle size. This study demonstrates how a uniform property assumption can cause erroneous derived thermal modeling, particularly over surfaces with a high percentage of boulders and large particle sizes. Future thermophysical studies of Earth or other planetary surfaces can greatly benefit from a multi-sensor approach combined with a higher spatial resolution visible dataset.