19-Nov-2015 -- I received a notification about Ross Finlayson’s visit to this confluence that my wife and I had visited almost 11 years ago.
After reading that he had lost his DJI Phantom quadcopter, I knew that I was the closest and only person crazy enough to try and recover it for him.
This confluence is the third closest to my house and I was very familiar with the area. It is fairly easy to get to, is within the Coronado National Forest, and you can drive to within about 500 meters of the point.
How hard can finding a bright white plastic object with red stripes in the middle of pristine desert be? This would be just like an outdoor version of “Where’s Waldo” except I’d only be looking for an object the size of his white and red striped hat. It turns out that it is actually much harder than it seems when you consider the slope and vegetation in the area.
Flying over the area would be the easiest way to quickly locate it. I was able to convince Peter, a flight instructor colleague of mine, that this would be a worthy and fun adventure. It took a while before our schedules synchronized but on 7 November we were airborne and heading east towards Dragoon. It was fun getting back to basics in a single-engine reciprocating aircraft with only a 500 foot per minute rate of climb!
Based upon Ross’ narrative I knew where he launched and the destination. If you draw a straight line between the two and add a suitable buffer on either side, you have a defined primary search area. I knew there was not line of sight between the launch and the confluence and also knew that the DJI logic included return to launch (home) if the signal was lost. My initial thought was this could be a clobber incident – where the air vehicle runs into one of the ridges somewhere along the flight trajectory.
Like many of my adventures, an abundance of technology was involved: We were navigating visually but had equipment that exceeded the legacy aircraft’s native capabilities. Peter was using an iPad for situational awareness with electronic aeronautical charts. My trusty Garmin was onboard counting down the bearing and distance. Two GoPro cameras were set up to document our search – one looking forward and the other out the left side. A Canon Image Stabilized 70-200 mm zoom lens attached to a Canon DSLR capturing a few images per second at a resolution of 5184x3456 would be the primary sensor for post flight analysis to hopefully help locate the Phantom.
We entered a slow left hand orbit at 7500’ MSL around the confluence. This first orbit was for orientation and a visual search using binoculars out the open window on the left hand side. The second orbit was taking continuous images bracketing the search area at minimum zoom setting. The third pass would be a slow descending pass into the valley while taking a continuous line of pictures at maximum zoom. The sheer number of trees and tall bushes as well as open areas with tall grass prevented visual acquisition. My hope was that several of these pictures taken from varying oblique aspects would contain several pixels of something bright white and in the shape of a DJI.
Over 200 pictures were captured during our flight. Several hours were then spent analyzing these images and I discovered a bright white object with right angles that appeared to be man-made and was approximately the size of the quadcopter in the top of a tree at the bottom of the valley just over 100 meters away from the confluence.
I sent Ross an email and explained that I had sufficient confidence based on this object appearing in several different pictures from different azimuths to hike out there to see if my item of interest was his quadcopter.
On the morning of 15 November, there had been strong rains but the system was moving northeast and the trend looked like it would pass north of where I was headed. I set out under clearing skies towards Dragoon. I got off Interstate 10 at Dragoon Road Exit 318 and turned onto Lizard Lane just beyond Mile Marker 5. This dirt road splits to the southeast here underneath the powerlines and becomes a double track with multiple areas of severe erosion but was suitable for a 4WD Tahoe all the way until the tanks at the end of the road where I parked.
My plan was to hike up the drainage to my target tree. If I was unsuccessful, I would climb up on the eastern side of the valley and search with binoculars to the west. At the tree, I did not find the drone and thought that possibly sunlight on rocks below and on debarked branches could have created the bright shape I saw. I hiked up and down the east side sitting down to scan up and down the western slope of the valley. Then it started to rain. A drizzle at first made it uncomfortable but when it started to pour – it was time to abort this attempt. After I got back to the Interstate, there was nothing but blue sky – Rats!
Additional imagery analysis led to the identification of two other points of interest with varying levels of uncertainty. A late evening work schedule allowed a subsequent daytime, mid-week search mission on the 19th. My first goal on this day was to visit the confluence (I had been within scoring range on the 15th but did not document my visit with any pictures). I set up my Trimble GPS on the most distinctive and identifiable item near the confluence – a stump that can be seen in Picture 3. The confluence area was pretty much as I remembered it 11 years ago. Mount Graham is visible to the north and you can see across the valley to the east and up the valley to the south. Steep valley walls limit visibility to the west. I then climbed up and down the western slope searching with binoculars. I was hoping that a higher elevation would help if the quadcopter had landed in a tree. If it had landed in the tall grass, identification would be very difficult.
After about 4 hours of hiking and searching I was resigned to another unsuccessful hunt and was making my way back along the west end of the drainage when something white caught my eye on the other side. It was in a bush and, even with binoculars, it was indistinct. Located on the other side of a vertical drop-off, it was different enough for me to carefully work my way down into the drainage and then back track up to where the lines of position I had created crossed.
Success!!! Fist-Pump + Happy-Dance + Yeah shouted loudly! I had finally found it!
I tapped the custom message button on my SPOT GPS Messenger to relay to several people (including Ross) that I had finally found his lost DJI Phantom quadcopter!
The Phantom was found suspended well within a bush just over 200 meters from where it was launched. It had been there for 50 days. Based upon the depth of penetration into the layers of this bush, I think it descended ballistically and not under the return-to-home flight logic with a slow rate of descent, otherwise it would have stayed on top of the bush based on the size and density of the branches. A couple of the propellers were bent but otherwise it looked to be in very good shape!
I hiked back to my vehicle and eventually made my way back home and remained extremely motivated that I had recovered his Phantom!
Review of my GPS trails shows that I had passed within 10 meters of his Phantom twice while walking up and down the drainage on my first search mission! There is a saying that you will eventually find something in the last place you look. Further review of my pictures showed several containing a white object in this bush – if you only knew where to look.
After a few emails back and forth with Ross, he suggested that I take a look at his video. I was just about to box it up and put it in the mail to him and decided to take the memory card out and have a look. The video file was corrupt because it was not properly closed before the camera lost power. I did a Google Search and found that the DJI camera would properly close the file if it was subsequently powered back on. So I tried that: I pushed the power button, the camera and gimbal came alive, there was lots of flashing LEDs and beeping sounds going on and then the gimbal stabilized! I powered it down and discovered that I was now able to read the video file! Ross has also written his own software to repair these improperly closed files.
His video captures the launch followed by a quick translation to directly over the area where it was eventually found. It remained stationary for about 30 seconds over this point (perhaps this is where it thought home was due to a poor GPS signal at the initial launch?) before it began climbing towards the confluence. The camera orientation panned vertically to the horizontal over the confluence and then began a clockwise rotation from north capturing views in all directions before it hesitated while pointing to the west and abruptly translated southeast before rotating counterclockwise and tracking back to the north towards the launch point. Ross believes that at this point is when his low battery signal initiated the return home feature.
My hope was that the final seconds of the descent would be captured so I could use that footage at the start of a quick video I created **VIDEO LINK WILL BE INSERTED HERE WHEN READY** of this recovery adventure. Unfortunately, the video stopped several seconds before impact, almost directly above where it was recovered. Rats! Ross gave me permission to use parts of his video in the one I created.
As far as the confluence goes, Pathfinder Office post-processing of my GPS data (3592 positions) reveals that my Trimble GPS receiver placed on the large tree stump was 6.61 meters north of the actual confluence with a Horizontal Precision of 2.4 meters. I placed it on top of the stump because any future visitors are likely to readily identify and locate the stump in their attempt to close on the confluence.
What a truly outstanding and worthy confluence/search/recovery adventure!!!
Now I *really* need to get my own DJI quadcopter!!