Keeping Drone Off the Shelf: How Iowa State Keeps Crop Research in the Air
At Iowa State University's Remote Sensing and Imaging Lab (RSIL), most of a flight day is ready before the team ever reaches the field. Both aircraft ride in the lab's field vehicle, where controllers and batteries charge off an inverter during the drive, and the sensors are already connected to a computer.
At Iowa State University's Remote Sensing and Imaging Lab (RSIL), most of a flight day is ready before the team ever reaches the field. Both aircraft ride in the lab's field vehicle, where controllers and batteries charge off an inverter during the drive, and the sensors are already connected to a computer. Most sites sit within 15 to 20 miles of Ames, and even the farthest are only about two and a half hours away, so the team can fly as soon as it arrives. The lab works this way from mid-April to mid-October, flying about twice a week, and this season alone its single pilot has collected 6 terabytes of data.
Why the Lab Exists
The RSIL was founded to support advanced remote sensing adoption for agricultural researchers at Iowa State University. Many research programs would not be able or willing to invest heavily to purchase expensive equipment and have a dedicated person trained in order to collect imagery from research plots. Collecting useful aerial data is not as simple as just flying a drone over a field. It requires specialized equipment, trained personnel, standardized workflows, and years of experience in data collection and processing in order to be efficient. By centralizing those resources, the lab helps researchers focus on their science while ensuring consistent, high-quality imagery is collected throughout the season, even for projects that require repeated observations over multiple years.
It gained traction quicker than expected because the team kept seeing opportunities to better connect researchers, specialized expertise, and high-cost equipment across the university. RSIL grew out of conversations between Fernando Marcos, a research scientist at Iowa State, and colleagues who recognized that many research groups faced similar challenges in adopting remote sensing technologies. Rather than each lab independently building and maintaining its own drone program, they saw an opportunity to create a collaborative resource that could share equipment, standardize workflows, preserve expertise, and make advanced aerial data collection more accessible to researchers across disciplines.
Many of those researchers are agronomists and plant scientists rather than remote sensing specialists, so when a graduate student graduated, much of the knowledge and many of the workflows they had developed left with them. The next student often had to spend months rebuilding those processes before they could start generating useful data. For many labs, that steep learning curve made it difficult to maintain a long-term remote sensing program. On top of that, traditional workflows could take three to four days before imagery was processed and ready for use, assuming the student was not busy with other field responsibilities, which is rarely the case during the growing season. By the time the data were available, opportunities to verify treatment applications, identify issues in field trials, or make timely management decisions could already have passed.
The RSIL addresses that by working as a support unit for researchers. It handles flights, processing, data delivery and training for more than ten research groups, so their teams get drone data without having to build that expertise in-house. Its reach also extends past Iowa, with partner units in other states flying their own fields and sending the imagery to the RSIL for processing for a multi-state soybean project.
Two Aircraft, One Decision
The lab flew DJI M200s for several years. As regulations grew less certain, it moved to NDAA-compliant aircraft and sensors, a decision tied directly to how its work is funded. "We rely primarily on grant funding, so when evaluating new equipment, regulatory compliance is an important consideration," Marcos said. The team also wanted a single manufacturer across aircraft sizes to keep training, maintenance, and operations consistent. The IF1200 joined the program in 2025, and the IF800 followed in May 2026.
With two aircraft on hand, the research protocol decides which one flies. When a question needs LiDAR and multispectral data together, such as plant height, the IF1200 carries both sensors at once. When a single sensor answers the question, the IF800 goes up instead. The IF800 is also the choice for larger fields because its batteries can be hot-swapped mid-mission, something the IF1200 can't do, so a 100-acre field flown at 100 feet takes about four battery swaps to complete.
The Payloads
Each sensor on the lab's aircraft answers a different research question. The Sentera 6X multispectral camera produces vegetation indices and plant counts, while the Mad Nadir Mapping MNMLS-2πOS1 LiDAR measures plant height and the 3D structure of corn and soybean plots. The TerraRad Tech PoLRa 5 passive microwave sensor, new to the lab this year, measures soil moisture. Marcos is candid that the team is still learning that sensor, which is newer territory than the spectral imaging they know well.
Integration took little effort. The Sentera camera is the only payload that communicates with the aircraft, while the LiDAR and microwave sensors only draw power, and both of those manufacturers supplied Smart Dovetail-compatible mounts. "Overall, the integration process was simpler than expected and required very little customization," Marcos said.
Flight approach
Each sensor gets its own flight plan. For LiDAR, the lab flies at 100 feet AGL and about 6 mph with 80% overlap, running lines parallel to the crop rows to improve canopy penetration and point cloud quality. That approach typically produces around 1,500 points/m². Multispectral missions fly higher and faster, at 200 feet and about 15 mph with the same 80% overlap, targeting a ground sample distance of about 2 cm/pixel.
The microwave sensor works differently. Rather than capturing imagery, it records a single measurement per footprint at 1.4 GHz in the L-band, a frequency well suited to soil moisture but one that comes with a wide 35° field of view. To bring each footprint down to the plot level, the team flies it at about 30 feet.
Most missions cover research plots of up to 5 acres in about 15 minutes, and larger surveys are planned to stay under 25 minutes per battery. The Sentera imagery is geotagged with RTK NTRIP corrections, while the LiDAR uses its own GNSS antenna and has its trajectory corrected with PPK after the flight using the state's NTRIP network. Missions are planned in QGroundControl on a Windows tablet in the field vehicle, which connects directly to the controller.
Out in the Field
On a productive day, one pilot flies up to 12 missions, limited mostly by daylight, and one soybean field was flown 26 times to track the crop from early vegetative growth (V2) to full maturity (R8). That pace only works if every flight is repeatable.
"Most of the traits we measure, such as canopy development, crop greenness, and ultimately grain yield, are part of a continuous process throughout the growing season," Marcos said. "Collecting data the same way every time is critical to producing reliable results."
Small efficiencies make that consistency possible. Saved flight plans in QGroundControl let the team repeat missions across dozens of flights, Sentera calibration images can be triggered straight from the Herelink controller, and the field vehicle’s charging station keeps at least two pairs of batteries ready by the time a flight lands.
After two seasons, Marcos credits the IF1200's stability and the IF800's hot-swap batteries, and maintenance has been straightforward without significant downtime. The one issue has been some IF800 batteries becoming unresponsive. IFT's support team diagnosed the issue, and the affected batteries are being replaced under warranty. "They have provided the reliability and stability we need to support a large and demanding research program," Marcos said.
Data by Morning
Each flight produces 20 to 25 GB of data, and during the growing season the lab routinely collects hundreds of gigabytes a week. On the drive back, the team transfers it in a few clicks and automatically sorts it by farm and researcher. When the lab started in 2021, producing a single map for one field took at least three to four days. Now everything is processed and available the next morning.
That speed matters in research. "Waiting an extra day or two to review the data can mean missing a critical window in a research experiment," Marcos said. Fast turnaround lets researchers check quality during the season, adjust protocols, and sometimes refly an area before the window closes. They pull vegetation indices, plant counts, plant height, and biomass estimates from a web platform on a phone or iPad, and they can follow a field's progression across every flight of the season.
The lab built much of that system itself as three in-house tools: RSILfield for data collection and transfer in the field, RSILpipeline for processing, and RSILinterface for visualizing and working with the data. Processing combines the open-source OpenDroneMap with the team's own tools. Time was the biggest constraint, and off-the-shelf tools didn't fit the workflow, so the team drew on its mix of agronomy, remote sensing, and software experience to build tools around how researchers actually work. In some cases, one set of images becomes many products that carry far more information than the orthomosaic alone. The team has been developing the system for more than two years and is still adding features.
Measuring the Whole Plot
A few years ago, one trial collected plant height by hand, and it took four students a full day. Flying the same field took about 45 minutes, and processing took about 45 more. Time is only part of the problem with hand measurement, though. A study might have 200 plots with roughly 1,500 plants each, so researchers can sample only a few plants per plot, and the measurements vary between people and even from day to day.
"Instead of measuring a small sample, we can characterize the entire plot and do so consistently across flights, fields, and growing seasons," Marcos said. LiDAR is also less dependent on lighting, so it can collect data in conditions that don't suit multispectral imaging. The same shift applies to biomass, which normally means cutting plants down. Remote sensing can estimate it without destructive sampling, and Marcos believes that also allows slightly smaller plots and better accuracy.
What's Next
The lab isn't recommending management changes to farmers yet, and its focus remains on research. The next step is adapting the platform for on-farm use by choosing the layers and analytics most useful to farmers and extension specialists. The team plans to start collecting and processing data internally this spring, then launch pilot projects with key partners in early summer 2027. The fleet is growing too, with another IF800 planned for this fall.
For other universities considering a program like this, Marcos advises building for efficiency from the start and leaning on people who have done it before. "Don't try to do everything on your own. The drone and remote sensing community is incredibly collaborative," he said. "If anyone is considering starting a similar program, please feel free to reach out to the Iowa State University RSIL."
Sidebar: By the Numbers
Metric; the | Value |
Data collected this season | 6 TB |
Data per flight | 20 to 25 GB |
Flights per day | Up to 12 |
Faculty supported | 10+ |
Flights over one soybean field | 26 (V2 to R8) |
Processing time | 3 to 4 days per map in 2021, vs the next morning now |
Sidebar: Equipment and Flight Settings
Item | Detail |
Aircraft | IF1200A (since 2025), IF800 (since May 2026) |
Payloads | Sentera 6X multispectral, Mad Nadir Mapping MNMLS-2πOS1 LiDAR, TerraRad Tech PoLRa 5 passive microwave sensor |
LiDAR missions | 100 ft AGL, ~6 mph, 80% overlap, ~1,500 pts/m² |
Multispectral missions | 200 ft AGL, ~15 mph, 80% overlap, ~2 cm GSD |
Microwave missions | ~30 ft AGL, 1.4 GHz L-band |
Positioning | RTK NTRIP (Sentera imagery), PPK (LiDAR) |
Flight planning | QGroundControl on a Windows tablet |
The RSIL Team
Fernando Marcos, Ph.D., Research Scientist III
Mark Licht, Ph.D., Associate Professor and Extension Cropping Systems Specialist
Natan Seraglio, Ph.D. Candidate
