The thermal infrared (TIR) region of the electromagnetic spectrum is mainly used by volcanologists to investigate both the thermal and compositional characteristics of volcanic processes and products. This is done using a suite of orbital and airborne sensors. Some sensors such as the high-temporal/low-spatial resolution Advanced Very High Resolution Radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer (MODIS) are used as monitoring tools to detect new thermal activity or the presence of volcanic ash plumes. Higher spatial resolution sensors such as the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) or the Landsat Thermal Infrared Sensor (TIRS) are used to map smaller-scale deposits and processes. A wide variety of work has been performed using these sensors, including investigation cooling rates, surface textures, and compositions. This work has been expanded upon at the University of Pittsburgh by creating new ground based instrumentation to obtain detailed multispectral TIR image datasets of eruptive processes. The goal is to use these new tools and datasets to capture scales of volcanic processes (such lava flows and volcanic ash columns) never before possible.
The instrumentation that has been developed use two FLIR broadband thermal infrared cameras paired with a series of TIR wavelength filters design to allow multispectral image data to be collected. There are two different systems designed for different application and portability.
The first and older prototype system was developed using a FLIR S40 camera, which acquires 240 x 360 pixel TIR images. This camera has an uncooled microbolometer, and is capable of obtaining moderately-high speed image data (30 Hz). The six bandpass filters were initially designed to be manually placed in front of the camera lens, to allow six multispectral images. The six filters were fabricated to replicate the ASTER and MODIS TIR channels as well as those proposed for future TIR orbital instruments, that address current NASA Surface Biology & Geology (SBG) requirements. These filters are housed in a motorized wheel assembly, which can rotate and a range or predetermined speeds depending on the process being observed. The standard rotation of once every 3.5 seconds is typically used allowing 4-5 images per filter to be captured and averaged to increase signal to noise (SNR).
Figure 1. The original multispectral thermal camera system in operation at Sakurajima volcano, Japan. The wheel design allows for camera stability, and the rate of rotation means near-coincident image sets at each wavelength (as well as a broadband temperature image) can be obtained of a target.
To date, the camera has collected a large quantity of field data from Hawaii, Guatemala and Japan (Figure 1) during field campaigns over the past 7 years. The data collected from this camera have been used both as a proof of concept, but also to collect image data of ash rich eruptions from both Fuego and Santiaguito volcanoes in Guatemala. Spectral analysis of this data is ongoing, as are updated calibration methods, so that isolation of emissivity in each bandpass filter image can be acquired more accurately.
A second, much smaller camera system has recently been developed called the Miniature Multispectral Thermal Camera (MMT-Cam) system, which has adapted the principals from the first prototype. The entire system is encased in a weatherproof TIR instrument housing (Figure 2). The system incorporates the same six filters paired with a FLIR A65 TIR camera. This camera has an uncooled VOX microbolometer detector and is capable of acquiring larger image sizes (640 x 512) than the S40 at high frequency (30 Hz). The instrument system was developed along with software to acquire, calibrate, atmospherically correct, and post-process the data automatically. The calibration of the system and development of the correction algorithms were formulated using laboratory blackbody observations and well-defined rock samples analysis (Figure 3). Additional testing of the MMT-Cam to validate the accuracy of data acquired of dynamic molten lava surfaces was conducted at the Syracuse University Lava Project (Figure 4). These experiments determined the lens/filters attenuation and internal instrument radiance so that the transmission attenuation of the lens, filters, and the view path within the housing.
Figure 2. New multispectral FLIR camera system within the housing as well as the computer/power housed in the weatherproof case. The FLIR housing has the IR front cover removed to show the filter wheel and six, 18 mm filters.
Figure 4. Figure 4: Fully calibrated TIR spectral radiance 11.35 microns image data and the emissivity spectra acquired during camera testing at a basaltic lava pour. [A] Initial phase of the pour. [B] Final extent of the pour. [C] The mean emissivity spectra and temperature derived from the spectral radiance data, with the error bars indicating the normalized standard deviation of the lava extent. The mean emissivity is 0.762 +/- 0.02 for the initial phase at 1410 +/- 32 K and 0.883 +/- 0.03 for the final extent at 920 +/- 22 K, a 15.9% increase in emissivity over 490 K of cooling. The multispectral data were unsaturated because the MMT-Cam was in low gain mode and more importantly, the bandpass filters significantly reduce the flux to the detector. However, the broadband data were saturated. The MMT-Cam was orientated at ~25 degrees from the normal incidence angle of the pour surface at a distance of 1.5 meters.
Both camera systems have now been fully calibrated and the testing validates the instrument calibration procedure using a robust and rigorous procedure that corrects the filter transmissivities and instrument effects. Both systems have been deployed to active volcanic systems to analyze spectroscopic variability in volcanic eruption processes and products (including lava flows and ash plumes). Future work will deploy these systems to multiple active volcanic regions around the world to analyze variability in radiant emissions and spectroscopic characteristics. Additionally, work is on-going to develop an upgraded version of the MMT-Cam to improve observations of volcanic gas emission (MMT-gasCam), which is achieved by increasing the number of spectral bands from six to twelve.
This work is funded by the NSF (1524011) and NASA (NNX15AQ72H and NNX15AU50G).