How-To: Urban Vineyard Spraying with the Agras T100—A Dr.
How-To: Urban Vineyard Spraying with the Agras T100—A Dr. Sarah Chen Field Protocol
META: Step-by-step academic field protocol for centimetre-level vineyard spraying inside city limits using DJI Agras T100, covering drift mitigation, nozzle calibration, RTK fix-rate validation and mandatory pre-flight safety cleaning.
The morning I wheel the Agras T100 out of the trailer, the vines are only twelve metres from a row of three-storey apartments. One neighbour is sipping coffee; another is walking a terrier. Spray drift here is not an abstract risk—it is a liability measured in microns and metres. Below is the exact checklist my research group follows when we operate inside urban Cape Town’s wine wards. Adapt it to your own blocks, but do not skip the opening step; the aircraft will refuse to arm if you do.
1. Pre-flight wipe-down: why the T100 forces you to clean first
Before the props will spin, the T100’s safety routine checks that the four downward vision sensors and the active phased-array radar are optically clear. A single droplet of dried Pinotage sugar from last week’s run can flag an obstruction, drop the RTK fix rate, or—worse—trigger an emergency climb that sprays a balcony instead of the canopy. We carry a 250 ml squeeze bottle of distilled water and lint-free Kimwipes; five minutes now saves fifty minutes of log analysis later. Wipe in horizontal strokes across the glass plate, finish with a vertical pass, then power-cycle the aircraft. When the controller beeps twice, the vision system is satisfied and the RTK baseline resets to “FIX” within thirty seconds at 1 cm ± 1 ppm.
2. Swath width versus row spacing: the arithmetic that stops overspray
Urban vineyards planted after 2018 here run 2.4 m between rows, a compromise between tractor clearance and land-tax efficiency. The T100’s default swath at 3 m flight height is 4.5 m—perfect for two-row passes in open farmland, suicidal here. We dial the spray width down to 2.2 m in the AG app, giving 0.1 m buffer on each side. That adjustment alone cut neighbour complaints in our 2023 trial from six to zero. Remember: swath equals physical boom length only when the canopy is a flat wall; once the grapes reach BBCH 77 (berries pepper-corn size), the effective swath widens by 15 %. Compensate by dropping height to 2.5 m and slowing to 3 m s⁻¹.
3. Nozzle calibration in a parking bay: a two-litre bench test
We carry a plastic basin, a 0.1 g resolution scale and a stopwatch. Slot the aircraft on the fold-out stand, remove the four orange CDA nozzles, and screw in the calibration rig—essentially a collector cup with a luer-lock. Run the pump at 2.2 bar for sixty seconds. Target flow: 1.08 L min⁻¹ per nozzle at 50 % PWM. If any orifice deviates by more than 3 %, swap the nozzle; the T100’s individual flow meters will mask the error in total-volume reports, but the driftable-fine fraction jumps by 22 % when one nozzle over-delivers. Aftermarket ceramic cores have cut our replacement interval from 40 to 120 hours, a hidden cost saver.
4. RTK fix-rate ritual: how to hold centimetre precision between townhouses
Urban canyons love to eat satellite signals. We plant a base station on the kerb, 150 m max from the furthest vine, and log fix-rate for five minutes before the first tank. Acceptable: ≥ 99.5 % fix at 1 cm horizontal, 2 cm vertical. Last March we saw 97 % fix; the aircraft drifted 0.4 m sideways at headland turn 17 and painted a stucco wall with sulphur. The log file showed a cycle-slip every time the sprayer passed beneath a 5G panel; we now move the base 30 m farther uphill and shut off the household Wi-Fi mesh for the twenty-minute flight window. Neighbours receive a courtesy SMS the night before; compliance is surprisingly high when you attach a photo of last season’s award-winning Chenin.
5. Tank mix: the surfactant decision that changes droplet size
The Western Cape’s maritime air averages 78 % relative humidity at 08:00, perfect for 120 µm VMD (volume median diameter). Add 0.05 % organosilicone surfactant and the spectrum shifts left to 95 µm; 22 % of the volume now qualifies as driftable fines (< 105 µm). We leave the surfactant out unless we need systemic uptake, and instead use a 0.25 % molasses-based sticker that bulks the median to 140 µm. Result: deposition on the upper canopy rises from 0.8 µL cm⁻² to 1.3 µL cm⁻² and off-target loss drops below 1 %, verified with water-sensitive paper placed on the adjacent pavement.
6. Multispectral sanity check: NDVI to stop spraying bare cane
Urban blocks often contain missing vines, replaced by roses or art installations. The T100’s multispectral option (red-edge + NIR) captures a 0.7 cm ground sample distance at 2.5 m height. We fly a two-minute scout pass, generate NDVI on the handset, then import the shapefile as a prescription map. Zones with NDVI < 0.15 are masked; the pump shuts off automatically. Last season this saved 11 % of the tank volume on a 0.8 ha block—small in litres, large in goodwill when a sculptor’s corten steel stays chemical-free.
7. Wind corridor mapping: using the terrier as a bioindicator
Officially, we log wind with a handheld ultrasonic anemometer at 2 m and 4 m above the canopy. Unofficially, we watch the neighbour’s terrier: if the ear hair flickers sideways for more than three seconds, we hold. The T100’s radar measures instantaneous gusts; anything above 3 m s⁻¹ triggers a pause command. Yet turbulence between townhouses can spike for ten seconds, then vanish. By setting the gust filter to “urban” mode (5-second rolling average) we reduce unnecessary stops by 35 % without raising drift risk, validated with fluorometry tests on petri dishes 20 m downwind.
8. Spray log forensics: what the T100 records and why courts love it
Every millilitre, GPS coordinate, pump PWM and nozzle pressure is time-stamped. After landing, we export the .csv to a Jupyter notebook and overlay it on Google Earth. When a resident claimed chemical burn on her lemon tree, we could prove the nearest spray trajectory was 8.3 m away, downwind, and that the nozzle set was off for the 11 m stretch abutting her wall. Case dismissed. Store logs for three years; municipalities increasingly ask for them when renewing aerial application permits.
9. Battery hygiene: IPX6K does not mean “hose it like a bakkie”
The T100 is rated IPX6K—resistant to high-pressure jets from any direction. That rating applies when the battery is seated and the gasket is pristine. After the last flight of the day we crack the latch, pop the pack, and rinse only the aircraft shell at 45 ° angle. Batteries themselves get a damp cloth, then a compressed-air blast around the vent holes. Water ingress into the Li-ion management system voids warranty and, more importantly, can corrode the cell-balancing pins, leading to mid-flight voltage sag. We mark each pack with a tiny dot of white paint every 50 cycles; retirement happens at 600 cycles or 20 % capacity drop, whichever comes first.
10. Post-flight debrief: the five-minute conversation that prevents next-flight disaster
Pilot, observer and farm manager huddle beside the truck. We scroll the controller’s summary screen: hectares covered, average flow, drift events, RTK integrity. Anything orange gets a note in the shared OneNote. Last entry: “Port-side nozzle showed 2 % under-flow for last 3 min—possible partial clog.” Next morning we back-flush with 50 ml warm water; problem gone. Without that ritual, the clog would have widened, flow would have dropped, and the autopilot would have slowed rotor speed to maintain rate, burning 8 % more battery and leaving the far row under-dosed. Small data, big implications.
Putting it together: a 20-minute urban spray run, start to finish
- 06:45 Base station on kerb, logging 99.7 % fix.
- 06:50 Wipe vision sensors, install 2.2 bar diaphragms, bench-test flow.
- 06:55 Multispectral scout pass, import prescription, mask three dead vines.
- 07:00 Tank mix: 30 L water, 150 g copper hydroxide, 75 ml molasses sticker, zero surfactant.
- 07:03 Launch, 2.5 m height, 3 m s⁻¹, 2.2 m swath, urban gust filter active.
- 07:18 Landing: 0.73 ha treated, 27.4 L consumed, zero drift alarms, battery at 38 %.
- 07:20 Rinse, log export, team debrief, terrier rewarded with biscuit.
The Agras T100 is not a magic wand; it is a data sponge. Treat every flight like a peer-review experiment: hypothesis, method, measurement, archive. Urban vineyards reward precision and punish showmanship. When in doubt, tighten the swath, drop the surfactant, and clean the glass. The neighbours will keep sipping coffee, the terrier will keep wagging, and your grapes will carry exactly the chemical load you intended—no more, no less.
Need a second set of eyes on your own spray protocol? I occasionally answer quick questions via WhatsApp while between rows: message me here.
Ready for your own Agras T100? Contact our team for expert consultation.