Wave Blue

The Issue for Large Farms

An important issue in the agricultural sector is the time and resources it takes to manage pest control, weed control and crop stress monitoring over very large areas of land. More and more farmers are looking towards new technology such as Unmanned Aircraft Systems (UAS) to drastically reduce the time it takes to manage these issues, manage them more effectively and save resources. Rather than having growers evaluate fields manually on foot or by tractor, this technology allows farmers to gain immediate knowledge about the status of their fields in shorter periods of time. This information can now be gathered whenever and wherever it is needed, minimizing the response time required to address issues and maintain crops.

Winnemucca Farms is the largest single integrated farming operation in Nevada and one of the largest in the U.S. They have approximately 28,000 acres of land available for farming located across two valleys. The Winnemucca farm manager was using manned aircraft to acquire color infrared images of the fields. These images didn’t have the detailed spatial or spectral resolution that can be acquired from UAS with multispectral and standard color sensors.

The Desert Research Institute (DRI) and AboveGeo (a UAS technology company) met with the Winnemucca Farm Manager to discuss their current image products from manned aircraft image collection and what additional data would be of benefit to them. Winnemucca Farms typically purchases color infrared images as hard copy products, that do not have sufficient spatial resolution or overall image quality to provide the level of information for the management decisions. High quality images with fine spatial resolution would provide more detailed information on small areas of crop stress much earlier. Specifically, the farm manager stated that a one-time high spatial resolution set of images depicting elevation differences as shaded hillslope images would be helpful, but a more beneficial product would be the Normalized Difference Vegetation Index (NDVI from red and near infrared bands) image depicting crop health at critical times, as well as derivative estimates of total crop cover acres per field.

The Solution

The term “UAS” is preferred over “drone”, because “drone” is the term that was used in World War II for dummy unpiloted aircraft that were sent up into the air without much control, which were then shot at for target practice. UAS, unmanned aircraft systems, is a much more sophisticated platform with a suite of available sensors, so NASA and other organizations prefer using the term “UAS.” UAS can also be defined as Unmanned Autonomous System, which would refer to either ground and air based vehicles.

One of the great advantages of the UAS is the detailed imagery it provides, which is what was lacking at Winnemucca Farms. Because the system is flying low over the field and mosaicking very detailed images (e.g., cameras that have 20 mega pixel versus five megapixels), it allows you to capture much more detailed imagery. This is what the UAS brings as a solution. In addition to this, through computer-aided data analysis, DRI also created shaded hillslope imagery that provided a 3-D view of the elevation differences within the field. “This is something that the Winnemucca farm manager told us was of high value to him,” according to Lynn Fenstermaker, Biology Research Professor at DRI.

Several Nevada companies were approached (Alaska Aviation Proving Ground, Inc., VeraScan and AboveGeo, formerly AboveNV) and based on the study site location, company UAS fleet and availability, AboveGeo was selected to participate in this project.

What this technology does is it gives the farm manager an opportunity to fine tune the amount and placement of water, herbicides and/or pesticides so that they’re not applying the same rate to the entire field. This allows the farm managers to reduce costs by reducing the amount of pesticides, herbicides and water applied to each field and each of these applications have associated costs. In the case of water, if used in low spots where water is going accumulate, the amount of water applied to these areas can be reduced by either increasing the speed of the center pivot irrigation arm or decreasing the amount of water to be applied. The farmers can then preferentially apply water, herbicides and pesticides to the areas where it’s most needed and thereby cutting down on overall costs and conserve water.

Pilot Background

This was a NV Governor’s Office of Economic Development Knowledge Fund project that was jointly funded by WaterStart and the overall Knowledge Fund. The Desert Research Institute was interested in working with a couple of farms on the project, but only received one willing to participate, Winnemucca Farms. Winnemucca Farm’s chief administrative officer, Samuel Ralston, was at the time a member of the WaterStart board of directors and helped facilitate the project.

The specific goal was to conduct research to determine which platforms and sensors would be able to provide data of value to farm managers, with a particular focus on addressing drought, irrigation, water stress, weed and pest related issues. The project tested the capabilities of different sensors and different platforms such as fixed-wing vs. multirotor UAS and standard color versus multispectral cameras. The research was conducted primarily on alfalfa, peas and potato fields.

DRI tested the applicability of UAS data to address large-scale, multi-crop agricultural needs, particularly related to herbivory, salinity and water-related crop stress. The main goal was to use UAS data acquisition to identify and map agricultural crop stress that will lead to improved water use while maintaining and/or improving crop yields.

DRI provided imagery, made it available on a website as well as providing hard copies for the farm manager. According to DRI, they received a lot of support from Winnemucca Farms in this project regarding information needed on planting, when the crops were emerging and what were the best times of the growing season to capture imagery so that they could capture maximum crop stress. The majority of the deliverables in this project were to provide imagery, although there was one case in which DRI provided data sets regarding estimated value of crop loss due to herbivory. DRI estimated the area of lost or low yield and then the value of that crop loss.
DRI conducted the pilot project from June 9th 2016 through September 12th of 2017 and they conducted seven field trips where UAS imagery was acquired, four data acquisitions in 2016 and three in 2017.

Pilot Results

Multiple UAS flights were planned and completed during the 2016 and 2017 growing season at Winnemucca Farms. Flight acquisition dates included the following for each year:
 2016: June 9; July 7-8; August 11-12; and September 7
 2017: April 1, August 1 and September 12

For each date of UAS operations, imagery for the entire 130 acres of each field (selected by the farm manager) were acquired. All fields are round center-pivot irrigated fields and AboveGeo acquired imagery, both standard color and multispectral imagery over the entire field. DRI staff then mosaicked images for each field together into one large data set. They then conducted an analysis of the resulting mosaic image to look at variation in plant cover and correlating plant cover to what caused variations in cover.

“I don’t know that we gave them everything that they were hoping for because the project would have required more involvement of the farm management with the DRI team,” according to Lynn Fenstermaker, Research Professor and Deputy Director Division of Earth and Ecosystem Sciences at DRI. According to Lynn, at the end of the project the farm manager, “went 180 degrees from being reticent about the project to being very excited about the products and wanting to use UAS in the future to look at every single field and determine how best to improve yield.” DRI conducted the pilot project for a subset of fields at the main farm (identified by the farm manager as high priority), which was a small subset of the entire Winnemucca Farm acreage.

The 3-D hill-slope images that revealed elevation variations within each field were seen as very valuable to the farm manager. The imagery showing percent vegetation cover right before harvest was also considered to be of high value because these images allow the manager to assess what the potential yield would have been if there would’ve been 100% cover. This type of imagery and image products are very helpful to for targeting pest management instead of blanket pesticide application across an entire field.

The pilot project did not have the time/resources to track specific metrics such as water saved, pesticides reduced etc. DRI talked to the farm manager about a variable rate irrigation application to reduce water use based on findings, but Winnemucca did not have that equipment available on the farm at that time. The type of equipment that would be needed would be specific computer software that is still under development and variable rate nozzles on every irrigation center pivot. While Winnemucca could alter the application of water on the field, they didn’t have the ability to alter at the fine level of detail that the UAS data would have recommended.

It was discovered during the course of the project that multi-rotor platforms, i.e., DJI Matrice (6-rotor) and DJI Phantom (4-rotor), provided greater stability, particularly during windy conditions. GPS integration was found to be very important for providing the best possible mosaic of individual images into a composite image for an entire field. Another conclusion from image analysis results is that the better the camera spatial resolution (i.e., higher megapixel rating) the better the mosaicked image products.

It was also determined through the testing of several different software programs that a combination of AgiSoft PhotoScan and ArcGIS software provided the best set of image processing and geospatial tools for developing the final composite image products as well as the vegetation index images that the Winnemucca Farms farm manager identified as being most beneficial for crop management.

DRI looked at different sensors and found that high spatial resolution was more important than spectral resolution, which is not what they had anticipated. They also found that with the winds that tend to come up midday in the Mojave and Great Basin deserts, using a multi-rotor platform actually worked better than a fixed-wing platform. The fixed-wing has more surface area which provides greater 3-D movement of the aircraft, which in turn, makes it more difficult to mosaic images due to the variable orientation (3-D tilt) of individual images. Whereas that multi-rotor is more stable in the wind, more level and delivers better quality imagery.

The final significant accomplishment of this project was the development of a user-friendly, web-based visualization tool that would be intuitive for farm personnel use. DRI developed the online tool using software available at the Institute, including the resulting ArcGIS image products. The webpage interface was developed using Adobe Flex integrated with ESRI?s ArcGIS Server version 10.5.1. A MicroSoft SQL Server 2012 database was used to store and host the UAS image products for Winnemucca Farms. Scalable data layers available from ESRI?s public cloud services were used for the continuous image background and infrastructure features, and the individual UAS image products were published to the site. Individual UAS image products can be viewed using drop down menus that allow farm personnel to select the image type or types they would like to view for respective fields and dates of acquisition. The website has capabilities that will allow the farm manager to zoom into high spatial resolution UAS images of each center pivot irrigation field.

Lessons Learned

A final report published by DRI and AboveGeo listed the following lessons learned from the project:

  1. It takes time to develop trust with Farm Managers given the commodities risk associated with large corporate farms. Non-disclosure or similar informal agreements to limit access to data and results to only farm employees are good first steps in gaining trust.
  2. Initially it was thought that fixed wing platforms would provide the best data given their ability to cover large areas with the least number of landings to replace or recharge batteries. Because wind is almost always an issue and the fixed wing aircrafts tested provided imagery at too many orientations for accurate georectification and mosaicking, they discovered that rotor UAS provided better quality imagery.
  3. Sensor spatial resolution is as important if not more important than spectral bands to produce the best image products.
  4. AgiSoft PhotoScan appeared to perform as well as Pix 4D software, which is a more expensive commercially available software package for processing UAS images. Web-based data processing software, such as Drone Deploy, may be a good alternative option.
  5. Image processing is the most time intensive aspect for commercial UAS businesses given the large volume of images acquired during a single UAS flight. At this time, there is not a cost effective solution available that can significantly reduce processing time, other than acquiring and/or leasing expensive parallel processing hardware and software systems or purchasing licenses for cloud-based image processing.

Further Adoption

This was a public-private partnership and DRI was partnering with AboveGeo on the project. Winnemucca expressed interest in contracting AboveGeo after the project to continue to fly UAS systems for their fields. They expressed interest in using the technology to cover all fields in the future but it has not been confirmed whether they have moved forward with this or not.

About AboveGeo

AboveGeo collects, analyzes and displays, aerial acquired, geospatial data using Unmanned Autonomous Vehicles (UAVs). AboveGeo provides services, equipment, and consulting to public and private sector customers – faster and for a fraction of the cost than can be done by existing means.

Translate »