top of page

Going Underground

This article reprinted from the NASAR publication "Response".

Thanks to owner Dave Cale’s ongoing generosity toward the Pittsburgh Grotto caving club and toward the cave rescue community, as well as help from the staff at Laurel Caverns, Pennsylvania (, we had the opportunity to conduct a dog-search exercise in the cave shortly after it closed to the public for the season.

Dennis Melko, a long-time grotto member trained in cave rescue, joined dog handlers from the Mountaineer Area Rescue Group (MARG) of Morgantown, WV to test our dogs’ ability to find practice subjects hidden in the cave.

We tested two different types of search-and-rescue (SAR) dogs in cave:

● Trailing dogs, which follow the the subject’s walking path by smelling scent that falls to the ground

● Airscent dogs, which follow scent carried by air currents directly from the subject and so do not follow the subject’s path (more on this later)

The initial setup of the exercise was a little different than we’d planned. We’d originally thought we’d stage in the Dining room, to the extreme northwest of the map below, and carry out search tasks in the unlit wild cave. When we showed up, though, the staff asked us whether we wanted the lights in the developed portion of the cave (what’s shown on the map) on or off — turning them off hadn’t even occurred to me.

Laurel Caverns consists of three different cave types. The upper part of the developed cave, at the southeast of the map, has been extensively modified for commercial tours to open crawling-only passages to walking and even out the floors. This portion of the cave slopes very gradually downward to the northwest and is very maze-like. The lower part of the developed cave, to the NW of the maze, consists of three roughly parallel and more sharply sloping passages that have been modified less extensively, to even out floors and install staircases (not shown on the map). Laurel Caverns offers guided tours of the developed cave portions. The “wild cave,” which lies off the map and farther to the NW, is largely unmodified and consists of much rougher cave passage with many rock “breakdown” piles resulting from rocks falling from the ceiling. Tours of the wild cave are available through appointment only, and require more strenuous effort and helmets and lights. Using the maze cave without the lights was a great idea because it allowed us to search the maze with our headlamps only as well as to stage the exercise from the cave office building rather than the Dining Room, which really simplified things and made the staging area warmer! Obviously, the post-public-closure date made this possible, and let me tell you the maze is very, very neat with the lights out.

Phase 1: Trailing Teams

Trailing dogs follow the scent of a specific person on the ground, so we give them a scent article that has their smell on it. Because I was the subject for the two trailing tasks in this exercise, they used a pair of gloves from my winter jacket to give my scent to the dogs. Trailing dogs, because they’re following a specific person’s scent, are somewhat resistant to contaminating scent from other people; but since we were starting out in our new series of tests of dogs in-cave, and because it made it much simpler to plan, we ran the trailing dogs first. Likely there was some residual scent from earlier visitors and staff, but my trail would have been clear and strong.

You can see my track in the cave in the map below. The light blue portion was the path I walked to hide for the first dog, stopping to hide a little above the staircase leading down the Grand Canyon. Later, from that location I followed the orange path downhill, hopping between the Grand Canyon and Devil’s Staircase then back up the Hall of the Mountain King, ducking into the side passage to the right (southwest) just downhill of the old cave entrance. As I walked, I left flagging-tape markers of my passage, a standard precaution in training to ensure the dog is staying on the proper trail. (When trailing teams test for certification, they do it blind, without such markers.)

Note that we didn’t have great maps that day, and so tracking our progress was a little dicey and the paths you see are estimated. (The map you see here, while out of date, is of a higher quality than what we were working with and will likely be our base map going forward.) Another interesting thing to note: The second hiding spot, being close to an opening to the surface, was fairly brightly lit and very cold that day. I was more comfortable in the first spot at the top of the Grand Canyon, which was warmer. I knew they’d started looking for me when the lights went out, and had a very philosophical few minutes in the dark waiting for them. Cavers will understand!

Ellie, a young mixed-breed dog just starting her training, did the first, shorter trail with her handler MARG member Anne Russell, accompanied by seasoned handler and MARG member Emylee Clements. I’d expected that the trailing dogs would be strong in-cave, and I wasn’t disappointed.

Though Ellie is a newbie, she followed my trail like a pro (or so they tell me), and soon was coming around the corner to find me, leading her teammates in for her reward. After the find, they returned to the cave entrance and I continued around the circuit from the Grand Canyon to the Dining Room then back to the natural entrance.

The second dog, a pit-bull terrier named Quest, handled by Emylee, is an operational SAR dog with much more experience than Ellie; he also had a tougher problem, as the trail was about twice as long and now had contamination from Emylee and Anne in its first portion. A larger group, including Anne, Grotto and MARG member Heather Houlahan, and Dennis, accompanied them in a trail that followed my initial trail to the top of the Grand Canyon and then back around. Once again, Quest was very strong at this task, following me around a number of turns in the maze before the big loop down to the Dining Room and back.

Phase 2: Airscent Exercises

With the trailing dogs done with their tasks, we no longer had to worry about search subjects’ trails crossing and contaminating each other. Airscent dogs are in some ways more vulnerable to contamination because they don’t follow a specific person but home in on any human scent. But they try to find the strongest scent, which helps to reduce the effects of contamination. In any case, contamination is common in real searches and just something we have to deal with — see Charlie’s task below for more on this!

One of the things we wanted to do in this part of the training, beyond having the airscent dogs search for and find subjects, was to feel out how to use them in-cave. If we need to walk up every passage in a maze to get a dog find, for example, we’re not really improving much on what cavers could do just searching visually (though the scenario of a subject who’s fallen into a crack and isn’t visible or calling back remains a possible use for airscent dogs). We needed to see if we could figure out a way of approaching a maze that allowed a detection at a far enough distance that we couldn’t have gotten a visual find. For the airscent dogs, we simply hid subjects in the maze and had the handlers walk their dogs in to search for them until they detected a strong enough scent on the air currents moving through the cave. On the surface this is relatively simple, and handlers tend to search using some kind of set pattern, walking through the area until the dog catches enough scent to follow it in. In-cave, the air currents got a lot more complicated. There was a gentle uphill (toward the southeast) breeze coming up from the Grand Canyon, but once it hit the maze the air movement got a lot more twisty and unpredictable. In addition to this, the maze didn’t really allow us an easy way of doing a simple search pattern. We did several airscent tasks in-cave, but I’m going to report on the two I handled, because I can dip into my training logs for details.

In the first task, I decided to attack the maze by walking on its outskirts. The idea was to circle the outer, walkable part of the maze so that the dog had a chance of picking up scent from deeper in it and ranging in independently. My dog, a newbie English shepherd named Sam, wore a lit collar that provided them with enough light to move beyond the range of our headlamps. (For those of you keeping track of the dogs we’ve been caving with, Sam is Charlie’s daughter, Rosie’s granddaughter and Pip’s great-granddaughter.) Dennis walked along with us on this one.

The obvious choice would have been to skirt the maze in a clockwise direction, in each intersection picking the left-most walkable passage. This would have made my first circuit of the maze start at the uphill end, with the wind blowing up toward us and making the chances of a detection at a distance better. But that part of the maze has a lot of passages that hadn’t been excavated in the early development of the commercial cave and so pinch off; picking a route was a little complicated.

I decided instead to search counter-clockwise, picking the right passage and so starting with the path down to the top of the Grand Canyon (the red path in the image). The wind movement wouldn’t be ideal, but the navigation would be easier, particularly since, again, we didn’t have maps as good as the one above.

Rounding the far (southeast) part of the maze, Sam ranged ahead of us, up into the Avalanche Passage (if you know the cave, this is where Dave set up a sound-and-light system to play the Hallelujah Chorus). She ranged up into the breakdown, checking each of the side passages and then returned to me whining and bumped me — which is normally her find signal (position is marked on the map with a black arrow, “Sam refind”). My guess was that either the subject was just past one of the turns, or she was getting scent up there but wasn’t confident following it herself, so I carefully moved up past the breakdown, to avoid disturbing the lights and wiring, joining her near the top. She ranged into the first passage to the left (which went in considerably farther than I’d remembered when we filmed a similar placement with our German shepherd Mel many years ago), then returned and bumped me again. I followed her in to find Anne hiding quite some distance into the crawl, and rewarded Sam for her good work.

In the next task, I worked with Sam’s mother, Charlie, a very experienced SAR dog. Dennis and Emylee walked with us.

This time, a little more confident, I used the left-hand-search method, rounding the (hopefully) downwind part of the maze. Again, the path (dark blue) is estimated, and choosing passages in this part of the cave was more challenging as I had to make quick decisions as to when a passageway was too pinched or too much of a storage space for equipment to use. (Again, early days — as we progress we’ll push pinches a little more.) Charlie, with her usual ton of confidence, ranged out of sight earlier in the task. Soon I heard her whining, out of sight ahead. Since if she’d made a find I’d have expected her to return and signal, I decided to move toward the sound. I found her moving up and down the Avalanche Passage. She’d found the residual scent from Sam’s earlier task, and was whining out of frustration that there was nobody up there. I encouraged her to continue, and we did. Soon after, rounding somewhere near the southeast edge of the maze, Charlie dashed into a passageway to the left, returning almost immediately with a bark signal (black arrow, “Charlie refind”). We rounded the corner, and Dennis pointed out Anne sitting in an elevated nook at the left side of the passageway. Charlie had found her, but only now saw how to get up to where she was; I waited for Chuck to join her, then return to me, barking again, and I walked the few steps in to reward Charlie.

Parting Thoughts

We ran six tasks that day with six dogs and four handlers, all of which were successful in terms of the dogs getting finds. Of course, we were after more than that; we did make good steps forward in understanding better how we could use the dogs in-cave.

As I’d expected, the trailing dogs have a clear and powerful use in-cave and I expect we’ll continue to develop this as a response capability (instead of just something we’re experimenting with or a novelty problem to make the dogs more flexible). That much seems clear.

We’re still learning about deploying the airscent dogs in-cave. All of the airscent tasks were successful; Sam’s was particularly so, since she pulled us into a passage that we might well have missed had we been searching visually. Charlie’s work was fine, but the placement of that subject didn’t give us much of a chance for a long-distance find (assuming the breeze was blowing uphill at this point; we need to keep track of air currents more carefully in the future). Another note for future training will be to place subjects closer to the middle of mazes, so that we can search from the perimeter and see if those long-distance finds are possible from the edges. Again, if we have to go too far into the maze it won’t be clear we’re improving over what human searchers can do. But in all I’m very happy with how the training played out.

Another scenario we’ll likely train for is that subject-down-a-crevice one. Laurel’s wild cave has some nice cracks and lower routes that allow you to hide people below the normal walking passage, but I’m also interested in hearing ideas for wild caves that can fit that bill. Unfortunately, Dave was unable to join us this time. Hopefully next time we train in Laurel (maybe in the fall after they close again?) he’ll be able to observe, as his cave expertise is profound and any advice he has would be invaluable (also he’d just enjoy it). We’d also like to get more experienced cavers, particularly folks strong in reading cave maps, so we can track our progress in-cave more accurately. Looking at the map, the dark blue and red paths aren’t really as far apart as I’d have liked (in other words I wasn’t doing a good enough job at finding the outer edge of the maze as we moved), and better way-finding as I worked the dog might help with that. (For those of you who know Heather and me and how strongly we feel about airscent dog handlers being able to navigate while working that may be surprising to hear -- but cave maps are a unique challenge and at least at first it winds up more like the limited navs that a trailing handler can manage while working.)

One wild cave I’d really like to try out is Bear Cave in Pennsylvania, because it has a fairly nice maze. I’ll check with the Mid Atlantic Karst Conservancy (, which owns the cave, closer to spring to find out when they’ll be OK with us working there. An added benefit is that, being a wild cave, we won’t have to pick a weekday to avoid crowds and so it may be a more convenient time for better attendance by folks outside our team. Stay tuned, I’ll be sharing more; going forward I’d like to do at least two cave trainings a year with the dogs. In addition to picking out caves that we know, we’re in contact with folks we know in the National Cave Rescue Commission, who have some other ideas that we may pursue in cooperation with NCRC and local caving grottos and SAR teams.

39 views0 comments

Recent Posts

See All


bottom of page