Thermal Infrared Remote Sensing of Active Basaltic Volcanoes: A Thermal and Spectral Deconvolution Approach

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) was launched in December 1999 as one of five instruments on the NASA Earth Observing System's (EOS) Terra satellite, and has proven effective for the detection and monitoring of volcanic eruptions and their associated products (Ramsey and Dehn, 2004). However, continuous advancement in analytical remote sensing techniques remains essential. For example, features associated with active volcanism commonly 1) are below the spatial resolution of the instruments 2) are more indicative of the state of volcanic unrest 3) tend to saturate thermal infrared (TIR) sensors due to their high thermal output. In addition, compositional, textural, and thermal heterogeneities can vary greatly within one 90 m TIR pixel, making accurate analysis and interpretations almost impossible without advanced techniques. Previous studies have shown that the radiance of an isothermal surface can mix linearly with respect to composition and texture, whereby emitted or reflected energy (TIR) from a heterogeneous surface is a combination of the radiance from each component proportionally to its areal percentage. However, where thermal mixing of a target's surface is involved, this technique is no longer valid, requiring alternative approaches to the solution. A thermal deconvolution algorithm has been developed to identify thermally mixed pixels and separate them into their hot and cool thermal components using archival and Urgent Request Protocol (URP) data from the higher spatial resolution shortwave (SWIR) bands of ASTER. These datasets targeted three active basaltic volcanoes exhibiting various thermal states including high-temperature lava flows with minimal SWIR saturation (Kilauea, Hawaii), low-temperature fumarole fields (Cerro Negro, Nicaragua), and high-temperature flows with significant SWIR saturation (Kliuchevskoi, Kamchatka). The results of this study show that this algorithm provides more accurate temperature estimates and corrections to the emissivity for better compositional mapping of the surface where SWIR radiance values do not approach minimum and maximum thresholds within each TIR pixel. This approach also serves as a rapid means for accurately identifying sub-pixel temperatures and minimizes the processing time, therefore allowing critical information to be quickly disseminated on these processes and hazards, which are commonly obscured in low to medium-spatial resolution orbital datasets