On the Orion-Eridanus Superbubble's feedback process, calibrating the NightHawk multi-spectral imager, and cloud/aerosol typing using a machine learning approach

First, clusters of the most massive stars generate a tremendous amount of energy throughout their lifetime. Their powerful stellar winds and radiation heat the material that surrounds them. When they die, they do so in one of the universe’s most spectacular displays: a supernova. When many of these stars live and die together in close proximity, the combined effect of their powerful winds and supernovae construct so-called superbubbles. Superbubbles have a shell of cool, dense material and an interior filled with million-degree, X-ray emitting plasma. We observed X-rays from the nearby Orion-Eridanus Superbubble to characterize the million-degree plasma in its interior. These observations and subsequent analysis are important because the energy output of groups of massive stars may play a role in heating the million-degree plasma that surrounds galaxies. High energy events within galaxies and their impact on their surroundings must be understood in order to understand how galaxies form and evolve over time. Our work indicated that stellar feedback can indeed heat the material that surrounds galaxies.

Next, wildfire activity is increasing around the globe with increasing impact on the environment and humanity. It is imperative to further develop tools we can use to monitor wildfire activity. To that end, we built and calibrated a remote sensing platform for observing wildfire activity at night. In comparison with some more complex, already established techniques, our device is simple and inexpensive. It uses four cameras, three dedicated to the blue, green, and red parts of the visible spectrum of light and one dedicated to infrared light just barely beyond what our eyes can perceive. Our device is effective at night, and using a simple technique, we can discriminate between artificial lights and light from wildfires. Calibrating the cameras turns them into energy measurement devices, rather than simple picture takers, and allows them to be used as scientific instruments.

Last, we developed a new technique for understanding measurements made by space-based lidar instruments. Lidars are similar to radars in that both instruments measure distances using light. Lidar uses a laser to do it, though. Lasers can also be polarized, which means we know that the light from the laser is only oscillating up and down, for example. When the laser light bounces off of stuff in the atmosphere, like clouds, smoke, or dust, the polarization of the returning light may change. We can use this information to determine what type of material the laser bounced off of. Typing material in the atmosphere using traditional methods involves an entire data processing pipeline, which is slow and has poor resolution due to spatial averaging. We developed a machine learning model that intakes raw data, circumventing the data processing pipeline, and determines what type of material the lidar has observed without sacrificing resolution.