Precision agriculture technologies have the potential to improve the efficiency and reduce the environmental impact of nutrient management. Five fields (two receiving manure) on three New York farms were intensively sampled to assess fertility patterns, showing good potential for variable application of lime, P, and K. High field-scale variability of P indicated environmental concerns in sections of some fields. Split-planter maize hybrid trials showed that variety response is site-specific and yield monitors are an effective research tool for fine-tuning hybrid selection. Field-scale N fertilizer strip trials showed that site-specific application of N fertilizer has little potential, but efficiency can be improved by annually adjusting N rates based on early-season weather. Digital georeferenced remote sensing images of bare soil and vegetation provide valuable information about variability in yield potential and crop stress in a field, but do not predict fertility patterns. Neither do soil maps. The utility of yield monitoring, global positioning systems, geographical information systems, and site-specific fertilizer spreaders were evaluated, and showed good potential for their use on farms. However, current technological challenges will likely prevent fast adoption of this technology. Educational efforts were made, including a two-day workshop, grower meetings, field days, extension in-service and CCA training sessions, new Cornell Guide recommendations for variable rate management, and several bulletin articles.
What is precision agriculture? We define it as the application of computerized data acquisition/control systems and information systems to land management. In that respect, crop production follows the manufacturing industry in incorporating computer-based technology into the production process, allowing for more efficient use of inputs and consistent product output. Precision agriculture is based on the premise that soil, crop, and pest-related processes are variable in space and time within fields. Information on these processes will, in principle, allow for more efficient crop production and environmental protection. Precision agriculture is strongly tied to several new or improved technologies, notably Global Positioning Systems (GPS), Geographical Information Systems (GIS), high-speed computers, and new sensors. Precision Agriculture can also build on new statistical procedures such as geostatistics and data mining methods.
Precision agriculture (PA) technology may involve many different types of equipment. Essential components are a Global Positioning System, an on-board (i.e., on the tractor or combine) computer, and a desktop computer with GIS software. The GPS is a device that allows you to determine your position on the land in terms of geographical coordinates. The unit typically does that by evaluating its position relative to a constellation of satellites and/or a land-based beacon. The GPS units generally sold for agricultural purposes are accurate within about 3 feet. It is an essential component of a precision agriculture system, because knowledge of the exact position in a field is needed to use the technology effectively.
The on-board computer can acquire data (e.g., from a yield monitor), or control an applicator (e.g., a fertilizer spreader). These tasks are performed while simultaneously evaluating the position of the field equipment using the GPS, i.e., the field data are georeferenced. The on-board unit is typically designed for just those tasks, hence the need for a desktop computer with Geographical Information System software to allow for the processing of such georeferenced field information.
The most widely adopted information gathering tool is the yield monitor which provides yield data for every 20 feet (or so) of combine travel distance. For most farmers, yield mapping is an entry point into precision agriculture. It provides good quantitative information on yield variability that may be used to make field management decisions. In addition, it provides an effective way to keep yield records and perform on-farm research (as collection of digital harvest data is automated).
For better fertilizer and lime application, intensive soil sampling is now being employed. This typically involves samples being taken within a field on a grid or by soil type. This information may then be used to apply these crop inputs at variable rates within a field. Hopefully, the increased costs of soil sampling and analysis are offset by more efficient use of the crop inputs.
There are several other information sources that are now being applied in precision agriculture. Notably, remote sensing has become more affordable. Services are now available to farmers that provide images from aircraft or satellite that can be used for a multitude of management decisions. Such information typically involves reflectance patterns in multiple spectral bands (including near-infrared, which is useful for assessing plant health), and is now available in digital, georeferenced format that allows for effective incorporation into a Geographical Information System.
Besides fertilizer and lime, several other crop inputs may be more efficiently managed with PA. Current or emerging technologies include variable seeding rates (e.g., higher plant densities in areas of higher yield potential), differential use of crop hybrids based on adaptability to variable field conditions, and variable herbicide application based on weed maps or on-board sensors. For the Northeast, the use of PA technology in manure application will be very valuable. Imagine in the near future having a record of the amount, timing, nutrient content, and location of manure being applied. This will provide important information that can assist farmers in more effectively using manure nutrients for crop production. Future manure spreaders may even be programmed to exclude applications in hydrologically-sensitive areas.
In the past years, we have experienced a tremendous leap forward in the development of hardware and software that allows for precision farming. However, the knowledge base to effectively apply precision agriculture technology is still underdeveloped. For example, if we want to use variable rate fertilizer application, we need to know how to sample the field, interpret the data and decide on how much fertilizer to put at each location. Similarly, we need to know how to effectively use a color-infrared image to make management decisions, or whether variety response and seeding density provide real benefits. The Cornell Precision Agriculture Initiative focuses on developing such a knowledge base and this SARE project was an important contribution to this effort, as it allowed for on-farm evaluations.
This project evaluated the use of precision agriculture technologies for the purpose of increasing the efficiency and reducing the environmental impact of nutrient management on New York farms, and to develop guidelines and education on the implementation of such technologies.
In 1999, field trials were initiated on five fields associated with three farms in the Central New York area: Elmer Richards and Sons Farms near Skaneateles, NY, and Split Pine Farms and Freier Farms near Geneva, NY. The former is a dairy operation, while the latter two are primarily cash crop farms. The fields range in size from 10 to 25 acres. Two of the fields regularly receive manure, and all were planted into maize in 1999, 2000 and 2001. Input was sought in February 1999 during a meeting of the New York Precision Agriculture Alliance (growers, ag business, researchers, educators and agency staff) to refine the experimental procedures for this project.
For each of the growing seasons 1999, 2000 and 2001, soil samples were collected on an non-aligned grid for each 0.5 acres. This oversampling allowed for spatial analysis of nutrient distributions and the determination of optimal sample spacing. Soil samples were submitted to the Cornell Nutrient Analysis Laboratory for standard soil test analysis. Soil samples were also collected for nitrate analysis at the same grid points in mid-June (PSNT), and at the end of the growing season (residual nitrate). In addition, pre-plant samples were taken at the beginning of the study in 1999. For each year, two corn hybrids (Pioneer 3752 and 37M81) were planted using a split-planter approach with strips along the length of each field. P and K fertilizer were applied according to Cornell recommendations. Starter fertilizer N was applied uniformly in each field at rates ranging from 25 to 60 lbs/ac. Supplemental sidedress N was applied in field-scale strips using two rates: 25 lbs/acre above and below, respectively, the Cornell-recommended rate. This allowed for an annual and field-scale evaluation of maize response to N fertilizer, and the need for upward or downward adjustment of those rates. For the years 2000 and 2001 , field-scale measurements were made on three of the five fields for maize growth, soil water content, soil strength and N content in the lower stalk.
For each field, digital georeferenced multi-spectral (color-infrared) images were acquired from aerial overflights at three (1999) or two (2000) times during the growing season. The first set of images for each growing season were obtained for evaluation of field variability under bare soil conditions, while the mid-season images were collected to assess crop health. The images were acquired and made available by Emerge-mPower3, and were processed to derive the SoilView and Vegetation Index (NDVI) products which aim at providing interpretive information on soil variability and crop health, respectively.
Yield data were collected using weigh wagons and digital georeferenced yield monitors. The yield monitors provide a spatial distribution of yield on the fields. Weigh wagons provided strip-average yields, which were used for calibration and validation of the yield monitors. The AgLinkTM software (courtesy of Agris) was used to analyze the strip trial data and evaluate the spatial distribution of hybrid and N fertilizer response. ArcViewTM (ESRI) and Vesper (Australian Center for Precision Agriculture) software were used for spatial analysis and presentation of the data.
A multitude of analyses were performed (most still in progress) on the various data, including the analysis of the spatial response to N and hybrid inputs using spatial difference maps in AgLinkTM, analyses of fertility patterns using geostatistics, correlation of fertility patterns with remote sensing and soils data, principal component analyses of remote sensing images, and various multivariate and data mining methods to determine data structures and correlations. In addition, composite maps were generated to illustrate field patterns.
- Data analyses and still in progress through the efforts of the principal investigators, one postdoctoral associate, and a graduate student, and expected to result in four or five refereed publications in the next year . The following preliminary results were derived:
Soil nutrient and pH variability on typical Central New York farms shows structured variability, thereby providing potential for more efficient use of nutrients through variable rate application. This is especially the case for the manured fields in this study. The potential benefits of variable lime application are high as pH variability on several fields was such that field-composited samples would have resulted in a severely biased pH estimates. P levels on the manured fields showed clear field trends which were agronomically and environmentally significant.
We determined that optimum sampling schemes for Central New York fields are based on a composited sample for every 3 to 5 acres. No clear correlations were observed between of soil fertility patterns with either soil maps or remote sensing images. Therefore, we are recommending that unaligned grid sampling is used if no prior information exists on field patterns.
PSNT results showed structured field-scale variability, suggesting the need for sampling of all parts of a field to obtain representative distributions, and a potential for the use of variable-rate N application. However, the yield data from the N-rate strips did not support variable N sidedress rates, i.e., N response generally did not vary over the field.
Precise management of N rates is critical for reducing nitrate leaching losses. Annual variations in optimum N rate, mostly due to early-season weather variations, requires seasonal seasonal management of N. Small adjustments in rates resulted in significant differences in residual N levels in soils. Based on concurrent research at the Musgrave Research Farm, we have concluded that precision management of N in the Northeast should focus on annual adjustment of N rates based on early-season weather conditions. We are working on developing a new recommendation system using real-time site-specific weather data and simulation models.
Yield monitors can be very useful for on-farm research. Besides fine-tuning N response, we determined this technology to be very useful for split-planter variety (hybrid) trials. On the manured fields, hybrid 37M81 outperformed hybrid 3752 by 7 bu/ac in 1999, 19 bu/ac in 2000 and 8 bu/ac in 2001. On the grain farms, little difference was observed in hybrid response, suggesting that the varieties interact strongly with locations (presumably soil quality and nutrient status). This indicates that on-farm research is useful and yield monitors readily provide a return on investment. A partial-budget spreadsheet was developed to analyze this for various-size farms. The yield monitor also quantifies yield losses in low-productive areas of the field, thereby providing information on the potential for remedial action.
Remotely sensed images can provide useful information for field management, and show considerable discrimination in field-scale soil variability and crop growth. Bare-soil and crop images were well correlated with yield potential, although they generally did not predict field-scale nutrient variability. Bare soil images readily discern hydrology and erosion patterns in fields and have potential use for identification of environmentally sensitive areas and fields and watersheds.
Precision agriculture technologies provide considerable potential for more efficient management of crop inputs, but limited user-friendliness of some technology components can pose barriers to their adoption. Considerable advances in ease of use and data exchange protocols are occurring to address these concerns.
The following publications have been developed:
· Two articles in What’s Cropping Up?, a newsletter for field crop management published by Cornell Cooperative Extension. One or two more articles are projected.
· One article in the Northeast Dairy Producers Newsletter
· Four or five refereed research journal articles are projected.
· A web site was developed for precision agriculture in New York, including copies of Powerpoint presentations (http://www.css.cornell.edu/research/precisionag/)
· A listserv was developed for the NYS Precision Agriculture Alliance
Three workshops on precision agriculture were held as part of this grant.
February 1999 workshop that established the New York Precision Agriculture Alliance and was used to solicit farmer involvement in the project.
Precision Agriculture Workshop and Meeting of the NY Precision Agriculture Alliance (December 14, 2000 and 65 attendees). Provided training on precision agriculture technologies and research updates; guest speakers from Indiana and Pennsylvania; group discussions on future directions; total of 12 hours of meeting time.
Half-day workshop involving the research team and the farmer collaborators on the research results and interpretation of the data (December 2001).
The following promotional and educational activities have been carried out in support of this project.
Niagara Peninsula, 2000 Ontario Corn Days (120 attendees total). Two presentations on precision agriculture research, including this project.
2000 Cayuga County Corn Day (80 attendees). Presentation on precision agriculture research, including this project.
1999, 2000, and 2001 Musgrave Farm Field Day (200 attendees total). Experimental procedures and results from this project were presented in an oral and a poster presentation.
1999, 2000, and 2001 Cornell Cooperative Extension In-Service Training 30 agents attending at each event). Research update on precision agriculture research.
2000 and 2001 Northeast Certified Crop Advisor Advanced Training (30 CCA attending at each event). Precise N management.
1999 Mid-Atlantic Crop Management Conference (90 CCA’s attending). Precision agriculture.
2001 New York State Ag Business Association Tour. Stop at Doug Freier’s farm and discussion of results of this project.
2001 workshop on Next Generation Nutrient Management Tools (30 professionals attending), Cornell University.
Three presentations at 2001 and 2002 Tri-Society Meetings
Research seminars on precision agriculture at Dept. of Crop and Soil Sciences, Cornell Univ. (twice); Dept. of Horticultural Sciences, Cornell Univ.; Slovak Pedological Society, Bratislava, Slovakia; Workshop on Soil Physical Management, Sao Paulo, Brazil; HortResearch and Massey University, Palmerston North, New Zealand.
Impacts of Results/Outcomes
This project has provided knowledge and information that is essential to the use of the next-generation crop management tools. The opportunities and limitations of a number of precision agriculture methods have been determined and guidelines have been developed on soil sampling, improved nutrient and crop management, new on-farm research methods, the utility of various data sources, the cost and returns related to PA tools. Deeper insights have been gained into spatial and temporal processes affecting nutrients and crop growth on New York farms. Many members of the New York agricultural and environmental communities have learned about precision agriculture concepts and technologies, and its potential for better management of the landscape, and the New York Precision Agriculture Alliance was established through this grant. Cornell University and the farm cooperators have built capability in research and outreach related to these new crop management approaches, which is providing a basis for future development of sustainable management practices (this includes as of yet unsuccessful attempt to obtain funds from the USDA-IFAFS program). Cornell University has also developed formal guidelines for precision management of crop nutrients. This project has also trained one M.S. student and a Postdoctoral Associate.
Early in the project, a partial-budget analysis was performed on the purchase of yield monitors for the purpose of farm-specific variety selection. The analysis included three sizes of grain crop acreage. The fixed costs of the yield monitor purchase and the variable costs of the satellite correction subscriptions were balanced against potential increases in yield from better matching of varieties with farm-specific conditions. Limited yield gains were required, especially for larger farms (>500 acres), to justify the purchase of a yield monitor. These yield gains were well within the hybrid response measured in this study, indicating the high potential for the use of yield monitors for on-farm research.
Further economic analyses are still needed for data collected during the three growing seasons of this study. This is planned for the next year.
We can’t fully assess the adoption of precision agriculture technologies as a result of this project. The three collaborating farmers have been strengthened in their understanding of the use of precision technologies on their farm. They also gained valuable insights in fertility patterns, corn hybrid response, field variability, relationship between hydrology and yield, etc.
Adoption by other farmers is ongoing, although difficult to quantify. Many consultants and extension specialists are aware of the results of this project. Farmers are increasingly purchasing yield monitors, and fertilizer dealers are increasingly using variable rate technology. NRCS and SWCD personnel is interested in precision technologies for farm planning.
Areas needing additional study
Precision agriculture is a wide-open research and educational area. The potential for integration of PA technologies into farm nutrient management is considerable, especially considering the need for the development of farm plans as a result of CAFO legislation. Key new technologies are under development, or have recently become available. This includes forage crop yield monitors and precision manure applicators, which have broad applicability in the Northeast. Methods of data integration need to be developed, and linkages between PA and watershed management need to be strengthened. Immense educational needs exist to familiarize land management specialists with PA methodology.