After a recent trip to visit my family, I found myself bumped onto a 6AM flight from Monterey (MRY) to Dallas (DFW). I mostly read, but some of my time was spent glancing out the window. During one of such times, I noticed some metallic pipes that appeared to be in the middle of the desert, straddled by hills or mountains about an hour after takeoff. I was curious as to who put them there and what they were for, so I stared curiously for a while and then snapped a pair of photos.

The first photo captured the pipes, which appear as shiny dots in the distance. After glancing at this photo, I knew it would be close to useless: you can't even tell that the pipes are pipes, and there was hardly anything I might use to determine exactly what part of the world it depicted. So, I took another photo which captured some nearby agricultural lands, some of the mountains, and what looked like a river. These photos also captured the time this took place: 7:07 PDT.

During my layover, I wanted to see if I could find any information on the flight path. A quick glance led me to FlightAware, which looked like it had the information I had hoped to find. I bookmarked the page and boarded my connecting flight.

FlightAware gave me some decent sense of where I might have been. I figured I might want to operate on a more precise level. Even though I could scroll through the flight timeline to see that I would have been near the California-Arizona border, I didn't have good coordinates and the FlightAware map doesn't have free satellite imagery. I couldn't find an easy way to get at the latitude and longitude data, so I opened up the browser console.

Searching through the console for some particular information led me to discover a ~trackpollBootstrap~ variable, which looked like it had some of the more relevant information. I threw together a couple of functions to get the coordinates from five minutes before I took the photo to five after, and then to put it into a CSV format suitable for import into some GIS software.

  function getInterval() {
      var track = trackpollBootstrap.flights["ENY4424-1634043600-schedule-0103:0"].track;
      var photoTime = new Date(1634217266 * 1000); // start ts from fa
      photoTime.setMinutes(photoTime.getMinutes() + 53); // time of photo
      var startTime = new Date(photoTime);
      startTime.setMinutes(startTime.getMinutes()-5);
      var endTime = new Date(photoTime);
      endTime.setMinutes(endTime.getMinutes()+5);
      console.log(startTime, endTime);
      var interval = [];
      for(let i of track) {
	  let d = new Date(i.timestamp * 1000);
	  if(d > startTime && d < endTime) {
	      let c = {lat: i.coord[1], lon: i.coord[0]}
	      interval.push(c);
	  }
      }
      return interval;
  }

  function printInterval(interval) {
      s = "X, Y\n"
      for(let i of interval) {
	  s += i.lon + ", " + i.lat + "\n";
      }
      return s;
  }
	  

With this done, I glanced at Google Maps' satellite imagery near some of the coordinates, and quickly found the agricultural features and the river I was after. I found Google's interface a little tricky to work with, so I fired up QGIS and grabbed some relevant Sentinel-2 data (the combination of which is both free and much more flexible).

My first step was to plot the coordinates I got on top of a normal map. QGIS gives easy access to OpenStreetMaps' data. All of my efforts so far suggest that my pipes lie just northwest of the Havasu National Wildlife Refuge (where I suspect the plane was at the time of the photos), placing the pipes near the California-Nevada border just west of Arizona. I found a pair of relevant Sentinel-2 observations that cover my area of interest, taken only two days prior. So, for the time being, I just grabbed the relevant data for bands in the visible spectrum. With those, I could now easily get a (close-to) true-color satellite depiction of where the pipes must be, with relevant OpenStreetMap data along side, helping guide my sense of direction and my intuitions.

Now I needed to put all this together to see if it would get me anywhere. I went back to the two photos. After some careful glancing, I found that the ridge in the lower right of the first photo cast the same shadow as the ridge in the upper left of the second! From this we can get a much better sense of the land, and where the pipes actually stand compared to the river and other easily found fixed features.

In the first photo, beyond the shared ridge, there are several smaller ridges, then the shining target points. Some of these smaller ridges are visible from the second photo, so I used the second photo to make a guess at a line from the ridge in the direction of the pipes. I found what looked to be the matching ridges in the satellite imagery, and followed this line past a few ridges, and found what I was looking for! Three unusual bright-grayish shapes in the desert, each isolated in dusty crop circles. These... don't appear to be pipes. I turned off the satellite imagery layer, I found, just west of I-15 south of the California-Nevada border, the Ivanpah Solar Power Facility.

Ivanpah is a solar thermal plant, which uses a vast array of mirrors to reflect sunlight towards one of three boiler towers, whose heat generates steam used to drive turbines. So the 'crop circles' are just vast arrays of mirrors, and the giant bright objects are so bright because all the reflected light is focused there. Ivanpah is apparently one of the largest solar thermal plants in the world, which merited some significant amount of press and even a visit from former president Bill Clinton.

So with a bit of data scraping, a bit of GIS, and some reasonable intuition, we now know that those mysterious bright structures out in the desert are bright for a reason. I am both surprised and delighted to find out that the unusual thing I saw from an airplane was this unique piece of infrastructure.