Expert Spraying with Neo 2: A Coastal Field Report from the Mangrove Edge

META: How the DJI Neo 2 handled salt spray, wind gusts and tight canopy gaps while dispensing biodegradable bird-repellent along a Hong Kong mangrove belt—real flight data, camera tricks and one cheap add-on that saved the day.

Chris Park, wetland-restoration contractor and sometime photographer, lowers the Neo 2 until its belly almost skims the incoming tide. From the shore we can’t even hear the props; only the soft click of the shutter reaches us through the salt air. In the controller’s live feed a grey heron stands ankle-deep, unaware it is being framed by a drone that is also—quietly—spraying a fine mist of grape-scented deterrent over the roosting trees behind it.

The job sounds simple: discourage migratory birds from two endangered Aegiceras corniculatum groves for three weeks while seedlings take hold. The site is a knife-thin sandbar at the mouth of a typhoon drain; every afternoon the sea breeze funnels between container-ship lanes and slams into the canopy at 38 km h⁻¹. No pilot on foot can weave through the prop-root maze without trampling saplings, and a boat’s wake would undercut the berm. A drone that can both film the work and carry a micro-sprayer is the obvious answer. The twist: the only unit available on 24 h notice is the 135 g Neo 2, a machine marketed to vloggers, not to people who wear neoprene boots for a living.

Why the Neo 2, not its bigger cousins

Weight is the first reason. Hong Kong’s Environmental Protection Department caps any unlicensed aerial pesticide platform at 250 g; the Neo 2 slips under that line with 115 g to spare. Second, the aircraft’s downward vision system doubles as a canopy altimeter, holding altitude to ±5 cm even when GPS drifts under the mangrove’s reflective leaves. Third—and this surprised us—the little props sit only 76 mm above the motor table, so when we invert the aircraft and hug the underside of a branch the wash fans outward, not downward, keeping droplet velocity low enough to avoid leaf burn.

Building a 38 g spray module that does not fry the gimbal

We start with a third-party accessory that rarely shows up in camera forums: the Ulanzi PT-11 triple-rail bracket, originally designed to hang LED panels off a phone rig. Machined from glass-fibre nylon, it weighs 17 g and its centre channel clamps perfectly around the Neo 2’s dorsal heat sink. On one side rail we mount a 10 ml polycarbonate perfume atomiser; on the other, a 3.7 V coreless pump scavenged from a broken facial mister. Total draw: 0.8 A for 90 s bursts. A silicone tube runs along the left landing foot, terminating in a 0.3 mm micro-nozzle aimed 30° down and 15° outward—exactly outside the prop arc. The atomiser is pressurised by a thumb pump before launch, so no onboard compressor is needed. With 6 ml of methyl-anthranilate solution the rig adds 38 g; take-off weight is still 173 g, 31 % below the limit.

Flight plan carved by the tide, not the map

High tide gives us 1.3 m of vertical clearance under the tallest Avicennia limbs; low tide exposes a spaghetti of pneumatophores that would trap a kayaker. We therefore fly on the flood, when branches bow to eye-level and saltwater muffles any downdraft rebound. The Neo 2’s 18-minute battery becomes 13 minutes once we account for pump surges, so we carve the grove into 40 m × 15 m strips, each flown in a single clockwise orbit at 2.4 m s⁻¹. ActiveTrack locks onto a fluorescent-orange frisbee we wedge in the mud as a reference point; the aircraft then autopilots the arc while Chris feathers spray windows with the custom C2 switch mapped to PWM channel 7. Obstacle avoidance stays on—sensitivity bumped to “slow” so the drone creeps around twigs instead of backing out and wasting prop wash.

Camera tricks borrowed from portrait class, not pilot school

The chinahpsy article that caught Chris’ eye last week argues that tension, not symmetry, holds a viewer’s gaze. One tip: place the subject at 4 % of frame area and let the eye hunt. We borrow the idea for progress documentation. Instead of the standard nadir survey shot, we drop to 1.1 m above the water and tilt the gimbal to −70 ° so the canopy ceiling fills the upper 80 % of frame; the sprayer’s nozzle enters at the lower right corner, barely 12 pixels tall. Viewers first see only leaves, then discover the tool at work—exactly the “slow reveal” the article praises. D-Log keeps the pre-sunrise colour temperature from blowing out the magenta in juvenile mangrove buds, while Hyperlapse at 2× speed compresses 6 ml of emission into a 12-second clip that looks like coastal fog rolling in.

Wind gust 38 km h⁻¹, obstacle gap 42 cm—data from the log

At 07:14 the anemometer on the berm spikes to 38 km h⁻¹; the Neo 2 leans 19° but still threads a natural portal only 42 cm wide between two prop roots. The flight log shows the forward vision camera detecting the left root at 0.68 m, triggering a 12 cm rightward shift; ground speed drops from 2.4 to 1.9 m s⁻¹ for 1.3 s, then recovers. No contact, no spray drift outside the target polygon. Those are the numbers that matter if you ever need to justify the aircraft to an environmental auditor.

The moment the accessory paid for itself

Mid-way through the third battery we hit a snag: the polycarbonate atomiser, chilled by evaporative cooling, threads loose and begins to pivot. A 2° tilt is enough to send droplets back onto the gimbal dome—bad news for the encoder. Chris thumbs the RTH button, but the tide is still rushing in; by the time the drone arrives the only landing spot is a 30 cm-wide oyster-encrusted buttress. We slap the Ulanzi bracket’s cold-shoe adapter onto a carbon-fibre monopod, extend it overhead, and land the Neo 2 vertically onto the makeshift helipad. Because the bracket keeps the spray rig 42 mm above the skids, the salt-coated monopod takes the drip, not the aircraft. Ten seconds of isopropyl on the carbon shaft, and we are airborne again. Without that third-party rail we would have scrubbed the mission and forfeited the morning tidal window.

Coverage metrics pulled from stills, not spectrometers

Back on shore we process 147 RAW frames into a 5 cm px⁻¹ ortho using DJI Terra. By overlaying the nozzle metadata (time-stamped on each image’s XMP header) we derive a deposition polygon that shows 87 % foliar hit inside the 1 m buffer from the trunk line—above the 75 % threshold the ecologist demanded. The remaining 13 % miss is concentrated on the seaward fringe, exactly where seedlings are sparse anyway. One flight, one dataset, one invoice-ready map.

What we would tweak next time

Swap the thumb pump for a 1 g CO₂ cartridge; pre-flight pressurisation is the longest chore.
Fly D-Log at ISO 200 instead of 100; the darker shadows hide noise when we pull stills for print.
Program a two-second pump off-delay at each orbit’s end to eliminate the last visible droplet streak.

A note on permits and public perception

Hong Kong allows sub-250 g aircraft without registration, but spraying anything—even grape flavouring—requires a pesticide-use exemption letter. We secured ours by presenting the obstacle-avoidance log and the 38 g payload declaration, proving the aircraft could not exceed 0.8 g m⁻² deposition at 2 m height. Having 4K footage that shows precise nozzle shut-off helped convince the committee we would not fog the hiking trail 60 m away.

Where to find the parts list (and a shortcut to advice)

If you’re kit-bashing a similar micro-spray rig, ping me for the wiring diagram and the 3-D printable nozzle shroud. I keep a WhatsApp thread for field engineers who trade notes on pumps, filters and salt-proof coatings: drop a line here. Replies can be sporadic when I’m waist-deep in mud, but every component we use is logged and bar-coded for traceability.

Ready for your own Neo 2? Contact our team for expert consultation.