Night Ice Navigation: How Drones, Radar, and AR Are Transforming Glacier Travel

The Khumbu Icefall has killed nearly 50 climbers since 1953. This chaotic labyrinth of shifting ice between Everest Base Camp and Camp 1 features house-sized seracs that collapse without warning and crevasses hidden beneath deceptively solid-looking snow bridges. For decades, the only defense was experience, intuition, and speed—move fast and hope the ice doesn’t move faster.

In April 2025, that equation changed. DJI FlyCart 30 drones began supporting the Icefall Doctors, the Sherpa teams tasked with fixing the route through this frozen maze. Using drone-based 3D mapping and aerial surveys, teams identified the safest path to Camp 1. For the first time in Everest history, no ladders or route-fixing equipment needed to be carried by humans through the Icefall. When a serac collapsed on April 15, destroying fixed ladders and widening crevasses, drones delivered replacement equipment within minutes.

This is just one piece of a broader technological revolution transforming how mountaineers navigate glaciated terrain—especially in darkness.

The Problem: Invisible Death on Ice

Crevasses are fractures in glacier ice that can extend tens of meters deep. Snow accumulates over these gaps, creating bridges that may support a person’s weight—or may not. During daylight, experienced climbers read subtle visual cues: slight depressions, texture variations, discoloration in the snow. The brain pattern-matches these signals against accumulated experience.

Night erases these advantages completely. Headlamps cast harsh shadows that flatten depth perception. Visual cues disappear. Every step becomes a calculated gamble.

Yet night travel is often unavoidable:

• Alpine starts at 2-3 AM catch firm, frozen snow before afternoon sun weakens bridges
• Summit pushes on 8,000-meter peaks require moving through darkness to hit weather windows
• Rescue operations don’t pause for sunrise
• Expedition logistics sometimes demand movement during optimal cold temperatures

The cruel trade-off: travel when snow conditions are safest but navigation is most dangerous.

Traditional mitigation strategies—roping up, probing with ice axes, following established wands—help but don’t eliminate risk. A probe test takes time, and time on a glacier means exposure to objective hazards. Fixed routes shift as glaciers move. Wands get buried or knocked over.

What mountaineers need is a way to see what human eyes cannot: the hidden architecture beneath the snow surface.

Ground Penetrating Radar: The Breakthrough Technology

The most significant advancement in crevasse detection isn’t thermal imaging—it’s Ground Penetrating Radar (GPR).

GPR works by transmitting electromagnetic pulses into the ground and measuring reflections from subsurface boundaries. When radar waves encounter a transition from snow to ice, or from ice to air (a crevasse void), the change in material properties creates a detectable reflection.

The Physics

Snow bridges over crevasses produce distinctive radar signatures:

• Dipping reflections above the void where the bridge sags
• High-amplitude reflections from ice layers at the snow-bridge base
• Slanting diffractions from near-vertical crevasse walls
• Absence bands where radar pulses pass through air rather than ice

New cracks and narrow crevasses (under 50cm width) show different signatures: no distinct snow bridge structure, fewer diffractions, but a characteristic band where pulse reflections disappear entirely.

Detection Depth and Range

Research across Antarctic and alpine glaciers has established GPR capabilities:

Frequency	Max Detection Depth	Snow Bridge Thickness
80 MHz	70+ meters	>30 meters
400 MHz	~40 meters	10-15 meters
800 MHz	~15 meters	<5 meters

Higher frequencies provide better resolution but less penetration. Lower frequencies see deeper but with coarser detail. Most mountaineering applications use 400-800 MHz systems that balance resolution with practical depth requirements.

From Research to Field Deployment

Early GPR crevasse detection required vehicle-mounted systems. Researchers on the Ross Ice Shelf attached 400 MHz transducers to 6.5-meter booms pushed ahead of tracked vehicles, profiling at speeds of 4.8-11.3 km/hr with real-time crevasse display.

The Strategic Crevasse Avoidance Team (SCAT) for the Greenland Inland Traverse combines:

• High-resolution satellite imagery analysis
• Vehicle-mounted GPR profiling
• Precise GPS waypoint logging
• Traditional mountaineering safety protocols

This integrated approach has documented and navigated around hundreds of crevasses per expedition. On one 10-km stretch, the team detected over 200 crevasses and classified each by snow bridge thickness and stability.

Drone-Mounted GPR: Taking Radar Airborne

The real game-changer is mounting GPR systems on drones, eliminating the need for humans to traverse dangerous terrain during initial reconnaissance.

The Technology Stack

A typical drone-GPR system includes:

• UAV platform: Multi-rotor or fixed-wing drone capable of carrying 2-5 kg payload
• GPR antenna: Custom-built lightweight antenna (often 80-400 MHz)
• RTK GPS: Centimeter-accurate positioning for precise profile localization
• Onboard processing: Real-time data logging and basic anomaly flagging

Real-World Results

A 2025 study at Switzerland’s Rhône Glacier demonstrated drone-based GPR’s capabilities. The UAV carried a custom 80 MHz antenna with RTK positioning, detecting structures tens of meters below the surface.

The data revealed dramatic changes over a single season:

• A subsurface cavity’s ice roof thinned from 9.6 meters to just 3.0 meters between July and October
• The cavity’s height expanded from 15.9 to 18.4 meters
• Meltwater channels formed and evolved beneath the surface

This time-series capability—monitoring the same features over days or weeks—is impossible with traditional ground surveys and invaluable for route planning on dynamic glaciers.

Advantages Over Ground-Based Systems

Factor	Ground GPR	Drone GPR
Operator safety	Moderate risk	Low risk
Coverage speed	5-10 km/hr	20-40 km/hr
Terrain access	Limited by surface conditions	Flies over obstacles
Repeated surveys	Labor-intensive	Automated flights
Setup time	Hours	Minutes

The trade-off: drone-based systems typically fly at altitude, requiring lower frequencies and sacrificing some resolution compared to surface-coupled antennas.

Thermal Imaging: Complementary Intelligence

While GPR detects subsurface voids directly, thermal imaging provides complementary information about snow bridge conditions.

The Physics

Snow bridges over crevasses have different thermal signatures than solid ice. The air gap beneath a bridge acts as insulation, causing surface temperatures to diverge from surrounding areas by as little as 0.1°C. These differences are invisible to human eyes but detectable with modern infrared sensors.

The Everest Implementation

In December 2025, researchers demonstrated integrated optical and thermal drone systems in the Peruvian Andes, monitoring glacial lakes and surrounding moraines. The DJI Matrice 4D series captured:

• High-resolution terrain maps
• Temperature variations across ice surfaces
• Meltwater flow patterns
• Thermal anomalies indicating subsurface changes

Traditional field monitoring was dangerous due to steep terrain and unstable slopes. Automated drone flights enabled frequent observations from safe distances, measuring glacier velocities, monitoring ice block acceleration, and detecting crack formation that might signal instability.

Limitations

Thermal imaging works best under specific conditions:

• Clear skies (precipitation obscures thermal signatures)
• Stable temperatures (rapid changes create noise)
• Sufficient temperature differential (midday sun can equalize temperatures)
• Dry snow (wet snow masks thermal anomalies)

Thermal provides probability estimates, not definitive crevasse detection. It’s most valuable when combined with GPR data or traditional probing to validate uncertain zones.

Machine Learning: Automating Detection

Processing GPR and thermal data manually is time-consuming and requires specialized expertise. Recent advances in machine learning are automating this analysis.

Neural Network Approaches

Researchers have applied Faster R-CNN (Region-based Convolutional Neural Networks) to GPR imagery, automatically detecting crevasses and continuous snow layers. The system:

• Processes raw GPR radargrams
• Identifies characteristic crevasse signatures
• Classifies features by type (deep crevasse, shallow crack, snow layer)
• Outputs confidence scores for each detection

Other approaches use Histogram of Oriented Gradients (HOG) features with Support Vector Machines (SVM), achieving reliable detection of both deep and shallow crevasses across varied glacier conditions.

From Detection to Navigation

The vision is end-to-end automation: drone surveys the route, AI processes the data, and results feed directly into navigation systems. “Safe tracks” monitored using GPR and differential GPS could be made available via download for use in handheld GPS devices.

Current systems aren’t fully autonomous—human verification remains essential—but the processing time from survey to actionable route guidance is shrinking from hours to minutes.

Smart Goggles and Heads-Up Displays

The final piece of the technology stack is getting information to climbers without requiring them to stop, remove gloves, and fumble with devices in harsh conditions.

The Smart Goggle Landscape in 2025

The smart eyewear market has exploded. Key products relevant to mountaineering include:

Julbo EVAD.2 Connected sports glasses designed for demanding athletes, featuring a heads-up display for real-time performance tracking while keeping eyes on the route. Originally designed for cycling and trail running, the technology translates directly to mountaineering applications.

REKKIE Smart Snow Goggles ($350) Transparent HUD technology displaying:

• GPS coordinates and altitude
• Speed and distance metrics
• Friend location tracking (915 MHz radio, ~2,000 ft range)
• Smartphone notifications
• Music controls

Battery life exceeds 10 hours—enough for a summit push.

PROVUU XR Goggles Using View-XR technology to enhance visibility in challenging conditions, from flat light on cloudy days to full whiteout conditions. The augmented view helps maintain spatial orientation when natural visual cues disappear.

Integration Vision

The endgame is integration: drone-generated route data overlaid directly onto the climber’s field of view.

Imagine: the drone completes its GPR survey. AI processes the radargram. Safe corridors appear as blue overlays in your goggles. Crevasse edges glow amber. Uncertain zones remain unmarked, prompting traditional probing before crossing.

The color-coding follows human factors research: warm colors signal caution, cool colors signal safety. Soft gradients preserve night vision adaptation. As conditions change—ice shifts, new cracks form—the overlay updates.

This isn’t science fiction. Each component exists. The integration work is ongoing, with several guiding operations testing prototype systems.

The Khumbu Revolution: 2025 Case Study

The 2025 Everest season provided the most dramatic demonstration of these technologies at scale.

Drone Operations

Airlift Technology’s DJI FlyCart 30 drones supported Icefall Doctors throughout the season:

• Payload capacity: 30 kg (66 lbs)
• Flight time: ~12 minutes per trip
• Traditional carry time: 6-8 hours for equivalent load

A single drone trip replaced a grueling day of exposure to avalanche risks and collapsing ice. Over the season, drones transported more than 2.5 tonnes of equipment and garbage across the Icefall.

Route Finding

For the first time, drone-based 3D mapping identified the safest route to Camp 1 before any human traversed it. When the April 15 serac collapse destroyed fixed infrastructure, drones delivered replacement ladders and ropes within minutes rather than requiring dangerous human carries.

Impact Metrics

Guiding companies report:

• 50% reduction in time and risk level for route fixing
• Zero ladder carries through the Icefall (historic first)
• Faster response to changing conditions
• Reduced Sherpa exposure to objective hazards

The SPCC (Sagarmatha Pollution Control Committee) also used drones to retrieve trash from high camps, demonstrating applications beyond safety.

Limitations and Required Discipline

Technology doesn’t eliminate alpine hazards. Understanding limitations is essential for safe deployment.

Environmental Constraints

Factor	Impact
Cold temperatures	Battery capacity drops 20-30%
High altitude	Thin air reduces drone lift capacity
Wind >25 km/hr	Sub-kilogram drones grounded
Precipitation	Obscures thermal signatures, risks electronics
Crevasse overfly	Drone altimeters may trigger dangerous altitude adjustments

When drones fly above large crevasses, onboard altimeters detect the sudden “drop” and may command the drone to descend—directly into the ice. Operators must set strict minimum/maximum height parameters and monitor for anomalous behavior.

Detection Limitations

False positives: Rock bands, refrozen melt features, and debris layers can trigger crevasse alerts. Teams learn to calibrate confidence thresholds through experience.

False negatives: Narrow cracks (<50 cm) and newly formed crevasses may not produce clear radar signatures. Traditional probing remains essential for uncertain zones.

Dynamic terrain: Glaciers move. A survey valid on Monday may miss features that opened by Wednesday. High-traffic routes require repeated monitoring.

Protocol Requirements

Successful deployment requires discipline:

1. Launch during weather windows—complete surveys before conditions deteriorate
2. Maintain backup methods—probing, GPS waypoints from daylight reconnaissance, established wands
3. Verify detections—technology augments human judgment, doesn’t replace it
4. Calibrate over time—experienced operators develop intuition for which confidence thresholds to trust

One guiding company’s mantra: “The drone tells you where to look carefully. It doesn’t tell you where to walk blindly.”

The Trajectory

As costs fall and systems become turnkey, the technology is spreading from elite expeditions to standard mountaineering practice.

Near-Term (2025-2027)

• Integration of drone survey data with smart goggle displays
• Lighter, more cold-tolerant drone platforms
• Improved ML models trained on diverse glacier types
• Standardized “safe track” databases for popular glaciated routes

Medium-Term (2027-2030)

• Real-time GPR from swarming micro-drones
• Predictive models incorporating weather, temperature, and historical movement data
• Regulatory frameworks for drone operations in wilderness areas
• Insurance and liability standards for technology-assisted guiding

Long-Term Vision

The ultimate goal: comprehensive glacier monitoring systems that track crevasse evolution over entire seasons, updating route recommendations automatically as conditions change. Climbers access current-day surveys via satellite download, overlay guidance appears in their goggles, and the traditional craft of glacier travel becomes augmented by ambient intelligence.

Mountains will remain serious places. Technology won’t eliminate risk—the forces involved are too vast, too unpredictable. But the balance is shifting. Careful climbers now have tools to stack odds in their favor when the route ahead disappears into darkness.

The Icefall Doctors of 2025 proved what’s possible. The next decade will show how far we can go.

---

TL;DR

• Ground Penetrating Radar (GPR) mounted on drones is the breakthrough technology for crevasse detection—seeing voids 70+ meters deep that thermal imaging and human eyes cannot
• 2025 Everest season: Drones transported 2.5 tonnes of equipment through the Khumbu Icefall, with zero human ladder carries for the first time in history
• Machine learning (Faster R-CNN, SVM) now automates crevasse detection from radar data, reducing analysis time from hours to minutes
• Smart goggles (REKKIE, Julbo EVAD.2) enable heads-up display integration—route guidance overlaid directly on the climber’s field of view
• Limitations remain: batteries degrade in cold, wind grounds small drones, false negatives still occur—technology augments but doesn’t replace traditional skills

Sources

• DJI Dock 3 Enhances Glacier Monitoring - Peruvian Andes
• Sports Illustrated: Drone Technology Improves Safety in the Khumbu Icefall
• CNN: Drones Deliver Supplies on Mount Everest
• Explorersweb: Drones Guide Sherpas Through the Khumbu Icefall
• The Cryosphere: 4D GPR Imaging of Glacier Collapse Features
• Cambridge Core: Drone-Based GPR for Alpine Glacier Surveying
• IEEE: Crevasse Detection Using Faster R-CNN
• REKKIE Smart Snow Goggles
• Julbo EVAD.2 Smart Sports Glasses