Atmospheric Phenomenon Research

Preparation starts weeks out, not the day before. In the lead-up to a multi-day session, EEG training intensifies — deliberate work through the NSE states to arrive in the field already in a stable, flow-adjacent baseline rather than trying to find it once out there. Data from prior sessions — flight paths, timing windows, scale estimates — gets archived and reviewed, and that record shapes where to post up and when to be watching most closely on a return trip.

Site selection isn't random. Vantage points are chosen for genuine sightline geometry — a location that can see across an entire canyon system and multiple ridgelines at once — combined with an instinct for dramatic, wide-open skylines shaped by years of photography before this project even existed. Camp is set up within a short hike of the chosen site: close enough to move quickly, far enough to stay quiet.

A field day runs long — typically nine to twelve hours, across two to four consecutive days per trip. The rule is simple: arrive with enough time buffer that time pressure is never a factor. Gear stays live throughout the day even during quiet stretches — camera, drone, and EEG headset all running periodically as tests, not just held in reserve. The EEG data doubles as NSE training data regardless of what the sky does that day, so no session goes to waste even on a quiet one. Pack-out happens with an hour of daylight to spare, every time.

Full equipment carried on a typical session: a dedicated video camera, a DJI Mini 3 drone, a Canon EOS camera, an iPhone, a Muse S EEG headband, an Apple Watch, a solar-powered battery pack for charging gear in the field, field notebooks, and a laptop for on-site review. A green laser is also carried as a contingency tool, though it hasn't been used to date.

Recording a sighting is only the first half of the process. Once footage comes back from the field, it goes through a structured breakdown to pin down what was actually recorded: size, speed, orientation, and distance.

The first pass draws on camera metadata alongside direct recall of the scene, where the camera was pointed, what else was visible in frame at the time. Distance and position get triangulated against fixed reference points in the shot itself: tree lines, ridgelines, and even atmospheric haze layers, cross-checked against the angle between the camera and the object. The full length of the clip gets tracked frame by frame, where the object enters and exits, how its apparent size changes across that span, and any spinning, morphing, or reorientation along the way.

Speed gets calculated by identifying a start and end point in the frame, mapping those points to real-world distance, then dividing by the elapsed time in the clip. That speed estimate then has to hold up against the size and clarity data from the same clip. If every measure agrees with the others, that's the signal the estimate is narrowing in on something real rather than compounding error from any single step.

From there, the case moves into Blender. The same real-world terrain gets modeled in 3D, and the estimated flight path gets built and tested against it directly, using animation and side-by-side comparison to further narrow size, speed, and trajectory. Flight paths and timing also get checked against the surrounding terrain and geography, looking for any correlation between where an object moved and the landscape it moved through.