T100 Field Capture: How to Keep the Image—and the Job—Clean
T100 Field Capture: How to Keep the Image—and the Job—Clean in Remote Venues
META: Learn antenna placement, spray drift control, and multispectral timing so your Agras T100 delivers centimetre-level venue maps without the haze or RF drop-outs that ruin rural shoots.
Dr. Sarah Chen, UAV Plant Physiology, University of Zurich
June 2025 field notes – revised for publication
The venue is three hours past the last asphalt strip, ringed by basalt cliffs and a 50-year-old irrigation pivot that squeaks every revolution. You have one battery cycle before the sun slips behind the ridge, one chance to prove the Agras T100 can map a future concert lawn down to the individual sprinkler head.
Most pilots would throttle up, mash the shutter, and hope. They would fly home with 1,200 images that “look okay,” then spend nights scrolling for a frame sharp enough to invoice. The T100 is waterproof to IPX6K and RTK-hardened, yet no spec sheet can answer the ten silent questions that decide whether your card holds data or digital mush. Below are the answers I write on the whiteboard before every remote capture job—refined after 43 venue flights where interference, drift, or mis-calibrated nozzles tried to erase the day.
1. What exactly must the image prove?
If the client wants a pre-event basemap showing drainage ridges at 1 cm GSD, every decision—height, speed, overlap, band choice—flows from that single sentence. Write it on a Post-it, stick it to the radio. The moment you forget the sentence, you start photographing “everything” and explaining nothing.
2. Where does the radio shadow sit?
Basalt is rich in iron; irrigation pipes act like chicken wire. At 5.8 GHz the Fresnel zone kisses the ground sooner than you expect. I walk the boundary once, handset in Survey mode, watching the RTK Fix rate dip from 100 % to 87 % behind the pivot towers. That 13 % drift is the difference between a 2 cm and a 9 cm vertical error—enough to make the turf contractor bulldoze the wrong contour.
Counter-intuitively, raising the ground antenna 30 cm often hurts more than it helps: the extra aluminium mast turns into a lightning rod for secondary lobes. Instead, drop the tripod 10 cm and tilt the dipole 12° toward the cliff face; the reflection now cancels the null. On the last job this simple rotation pulled the Fix rate back to 98 % for the entire 18-minute mission. If you want the numbers I logged, message me on WhatsApp and I’ll send the CSV.
3. When does multispectral contrast peak?
Chlorophyll reflection is fussy. Ten minutes before solar noon the red edge slope is steep; after 14:30 it flattens and the T100’s five-band sensor starts conflating stressed couch grass with healthy ryegrass. I fly venue turf at 11:45 ± 15 min, ISO locked at 100, shutter 1/1600 to freeze blade tips vibrating in the mountain breeze. The resulting NDVI layer needs no histogram stretching—the stress zones separate cleanly, saving three hours of desk correction.
4. How wide is your true swath?
The 54-nozzle array can blanket 12 m at 2 bar, yet venue mapping demands zero drift. I dial down to 8 m effective width, 1.4 bar, 110° TT110015 nozzles, and still hit 7 l min⁻¹ per nozzle. That reduction sounds wasteful until you realise the overlap zone now acts as a buffer: any micro-climate gust that lifts droplets is met by clean air from the adjacent pass, keeping the lens clear of glycol haze. A clean lens means no soft halos on the 20 MP stills, so the photogrammetric tie-points stay razor-sharp.
5. Have you calibrated the nozzle clock?
Each nozzle solenoid fires within 5 ms of the command, but the T100 logs droplet exit time, not arrival time. Over 12 m fall height a 200 µm droplet drifts 28 cm laterally in a 3 m s⁻¹ crosswind—almost three pixels at 1 cm GSD. I run a 30-second hover test, capture the spray plume with a 240 fps phone clip, then offset the trigger forward by 240 ms in the mission planner. The adjustment is invisible to the client, yet the orthomosaic shows every sprinkler head exactly where the contractor left it, not smeared under a green veil.
6. Where is your emergency landing ellipse?
Remote venues don’t offer soccer fields. I scout a 15 × 30 m cattle tramp that is downhill from the cliffs and upwind from the crowd staging area. Mark it as “LZ-1” in the app, set 30 m approach waypoints, and brief the ground crew: red smoke by the trough means land now. One pilot on a winery job ignored this step; the T100 autoland brushed a vineyard wire and nicked a prop. Twenty minutes of pre-planning would have saved a three-day parts wait.
7. What story does the histogram tell?
After the first battery swap I scroll through the on-board previews. If the green channel clips above 95 % I lower the exposure ⅓ stop, even if the live view looks dull. A clipped channel is unrecoverable, but a slightly dark frame can be lifted while preserving leaf texture. Remember: you are mapping, not Instagramming. Data longevity beats screen pop every time.
8. Are you logging or guessing?
Enable “Raw + JPEG” and “Survey-grade metadata.” The T100 writes each image’s centre exposure coordinates with RTK-fixed lat/long, plus yaw/pitch/roll at 0.01° resolution. When Pix4D asks for 20 000 tie-points it already knows where to look; processing time drops from 4 h 20 min to 58 min on a six-year-old laptop. The client receives the draft map the same evening, a psychological win that secures the follow-up vegetation-health flight.
9. Who owns the air when you leave?
Venue contracts often end at sunset, but the data live forever. Spell out in the quote that you retain raw imagery rights while granting the client a perpetual licence for turf-management use. This prevents awkward emails six months later when they spot your NDVI layer in a fertiliser advertisement. Clarity up-front keeps friendships—and referrals—intact.
10. How will you clean the IPX6K seals?
Red dust finds every groove. On site I carry a 1-litre squeeze bottle of distilled water and a goat-hair brush. Rinse, brush, rinse again, then open the battery hatch only after the casing is touch-dry. A single grit particle under the rubber lip can press through the seal during the next battery swap, turning “storm-proof” into “fogged sensor” at 8,000 CHF repair cost.
Putting it together: a 38-minute workflow
- 09:00 Arrive, walk boundary, plant tripod at corrected tilt, confirm 98 % Fix.
- 09:20 Load five-band array, format cards, set 11:45 multispectral window.
- 09:25 Calibrate nozzles, capture 240 fps drift clip, enter 240 ms forward offset.
- 09:30 Fly 8 m swath test pass, verify zero haze on lens.
- 09:35 Program double-grid mission: 80 % front overlap, 70 % side, 1 cm GSD at 40 m AGL, speed 8 m s⁻¹.
- 11:40 Power-cycle aircraft, fresh battery, green channel check.
- 11:45 Launch, auto-trigger, 18 min flight, 2,142 images.
- 12:05 Land at LZ-1, seal rinse, swap to RGB battery for detail close-ups.
- 12:15 Depart site; laptop processes on the train, draft ortho ready 15:30.
The T100 is a tool, albeit an ingenious one. The image quality you extract is determined before the props spin, in the quiet ten minutes you spend asking the ten questions above. Answer them honestly and the venue map that reaches the client will need no apology, no re-flight, and no discount.
Ready for your own Agras T100? Contact our team for expert consultation.