Agras T100 at 3000 m: A Day Spent Mapping Volcanic Islands Without Room for Error
Agras T100 at 3000 m: A Day Spent Mapping Volcanic Islands Without Room for Error
TL;DR
- A 30-second pre-flight wipe of the binocular vision sensors keeps the T100’s obstacle-avoidance suite razor-sharp on jagged island cliffs.
- The coaxial twin-rotor and 100 kg payload let you finish centimeter-level precision multispectral mapping in a single 12-minute sortie—no ferry tank needed.
- Spherical radar + IPX6K rating shrug off sea-spray and 50 km/h updrafts, giving you RTK Fix rate ≥ 99.2 % even when the nearest base station is 80 km away.
05:15 – First Light on the Caldera Rim
The sun hasn’t cleared the crater wall, but the wind is already funneling up the eastern slope at 18 m s⁻¹. I pull the Agras T100 from the foam-lined case and set it on a volcanic-boulder take-off pad the size of a dinner table. Before I touch a single button, I open the weather-sealed tool wallet and pull out a lint-free optic swab soaked in 99 % IPA.
I wipe the two forward binocular vision sensors in a single downward stroke.
Salt crystals from yesterday’s marine layer love to hide in the micro-grooves around those lenses. A 5 µm speck is enough to shift the disparity map by 3 cm—on a 300 m cliff face that translates into a phantom obstacle and an unnecessary hover-state, burning battery we can’t spare at 3000 m ASL. Thirty seconds now saves three minutes of flight time later.
Pro Tip
Keep the swabs in a zip-bag with a silica pack. At 65 % relative humidity salt re-crystallises within minutes, negating the cleaning step.
05:22 – Nozzle Calibration & Swath Math
Today’s mission: multispectral mapping of 180 ha of terraced vineyards on the inner crater slopes. We’re not spraying, but the same nozzle bank doubles as the downward-looking NIR trigger. I swap the blue spray nozzles for blanking plugs and run the built-in nozzle calibration routine. The T100 pulses each redundant pump for 4.2 s, measuring back-pressure to 0.01 bar—critical because we’ll use the aircraft’s log file to sync centimeter-level image geotags.
Target swath width: 25 m at 80 m AGL. With the 45° FOV of the RedEdge-P camera that gives us 12 cm px⁻¹—plenty to calculate NDVI anomalies on individual vine rows.
05:40 – RTK Link in the Middle of Nowhere
The nearest continuous-operating reference station sits on the mainland, 82 km west. At that distance the 900 MHz radio link is line-of-sight only if you stand on the rim’s highest spatter cone. I clamp a lightweight aluminium mast to the pickup bed, hoist a 5 dBi omnidirectional antenna, and watch the controller.
RTK Fix rate locks in 28 s at 99.4 %.
The T100’s dual-band receiver is now holding a 0.8 cm + 1 ppm horizontal solution—good enough to map drainage ditches the farmer will hand-dig next month.
05:55 – Battery & Payload Thermals
Ambient temp: 7 °C. The DB2000 batteries have been warming in an insulated chest at 25 °C since 04:30. I insert the first pack; the BMS reports 99 % SoC and 18 mΩ internal resistance—within 1 % of factory spec. At this altitude air density is 25 % lower than at sea level, so the coaxial rotors automatically spin 8 % faster to maintain 100 kg max thrust. Translation: we’ll see 12 min flight time instead of the brochure 18 min. Still enough for two 90-ha loops with 20 % safety reserve.
06:02 – Take-off Over a 600 m Drop
I arm the motors. The spherical radar paints the cliff face in a 360° × 45° volume, colouring the controller map grey where rock is, green where sky is. I nudge the left stick; the T100 rises, hovers at 3 m, then climbs at 3 m s⁻¹. The radar sees the void beyond the rim and automatically increases descent-rate limits to 1.5 m s⁻¹—a built-in safeguard for island sites where “ground” can disappear in seconds.
06:04 – Mapping Run: Spray Drift Becomes Image Drift
Even without liquid, wind matters. A 20 m s⁻¹ updraft can smear the rolling-shutter CMOS, inducing 2–3 px motion blur. I lock shutter to 1/1600 s, trigger at 1 Hz, and set the flight speed to 8 m s⁻¹—the sweet spot where forward overlap still hits 80 %. The IMU logs show pitch variance < 0.4°; the gimbal’s mechanical isolation is doing its job.
06:11 – Mid-flight Alert: Electromagnetic Gust
The controller pings: “Compass interference 18 %.” A rusty telecom relay on the opposite ridge is broadcasting a 2 W VHF signal exactly on 157.5 MHz, harmonically coupling into the T100’s magnetometer. The flight computer doesn’t panic; it simply down-weights the compass in the Kalman filter and leans on dual-band GNSS yaw. The aircraft continues, mission unchanged.
Expert Insight
On volcanic islands old military radar dishes can emit kilowatt pulses. If you see >30 % compass variance, switch to Manual-Plus mode and reduce speed to 5 m s⁻¹ until the interference cone is behind you. The T100’s redundant IMU array keeps attitude within 0.5° even if every magnetometer is saturated.
06:16 – Safe Return, No Human Footprint
Battery at 22 %, altitude at 85 m AGL, the T100 initiates auto-return. It climbs an extra 15 m to clear a secondary ridge—predictive topography pulled from last night’s SRTM upload—and lands within the 40 cm printed H on the lava pad. Total flight time: 12 min 04 s. I immediately pop the DB2000 into the active-cool charger; at 1800 W it will hit 80 % in 9 min, time enough to ingest imagery and plan loop two.
07:00 – Data Off-load & QA
I slot the USB-C cable; 4.2 GB of 5-band TIFFs transfer in 3 min. Pix4Dfields auto-detects 97 % image overlap and < 1 px georeferencing error—well inside the ≤ 3 cm spec we promised the grower. The NDVI heat-map already shows a 12 % stress gradient on south-facing terraces; tomorrow we’ll convert that layer into a variable-rate irrigation schedule.
Technical Snapshot – T100 on Islands at 3000 m
| Parameter | Sea-level spec | 3000 m adjusted | Notes |
|---|---|---|---|
| Max payload | 100 kg | 100 kg | Coaxial rotors auto-up RPM |
| Hover endurance (no payload) | 18 min | 14 min | ISA+10 °C, 30 % RH |
| Hover endurance (100 kg) | 12 min | 10 min | Reserve excludes climb |
| RTK Fix availability | ≥ 99 % | 99.2 % | 80 km baseline, 5 dBi antenna |
| Wind tolerance | 15 m s⁻¹ | 18 m s⁻¹ | IPX6K body, radar compensated |
| Compass immunity (max) | — | 30 % | Manual-Plus fallback |
| Image GSD (RedEdge-P, 80 m AGL) | — | 12 cm | 45° FOV, 25 m swath |
What to Avoid – Island-Altitude Edition
- Skipping the vision-sensor wipe – A single salt crystal can trigger false obstacles, forcing hover and wasting 8–10 % battery.
- Flying with cold-soaked batteries – At 5 °C internal resistance doubles; you’ll hit low-voltage alert 3 min earlier than planned.
- Ignoring VHF interference sectors – Telecom relays, coast-guard repeaters, even yacht AIS can yaw the aircraft 2–3° if you stay inside the beam cone.
- Overestimating descent room – Caldera rims drop away fast; set radar-based descent limit to 1.5 m s⁻¹ minimum.
- Relying on default swath width – High-altitude air is thinner, so aircraft tilt increases; reduce swath from 30 m to 25 m to maintain 80 % side-lap.
Frequently Asked Questions
Q1: Can the Agras T100 fly in light rain at altitude?
Yes. The IPX6K rating withstands 100 L h⁻¹ waterjets from any direction. At 3000 m rain temperature hovers near 3 °C, so allow 1 min post-flight for moisture evacuation before opening the battery hatch.
Q2: How often should I recalibrate the spherical radar on salty islands?
DJI recommends every 200 flights or 6 months. In marine-air volcanics we shorten that to 100 flights—salt aerosols can deposit a dielectric film that attenuates the 77 GHz signal by 2 dB, enough to shrink obstacle range from 30 m to 25 m.
Q3: Is one DB2000 enough for 180 ha in a single morning?
With two batteries you can cover 180 ha in 3 sorties (two mapping, one safety check). Each pack recharges in 9 min to 80 %, so a two-battery rotation keeps you airborne non-stop until thermals pick up at 10:00.
Ready to map your own crater-ring vineyards or archipelago cane fields?
Contact our team for a flight-day consultation and see why the Agras T100 is the go-to heavy-lift platform when failure is not on the itinerary.