There is a quiet, technical revolution occurring in agriculture that is going to impact the field of precision agriculture in the very near future. Lacking good terminology, I will call this revolution the “fusion” of machines, sensors and models. This fusion is being borne out of the explosion of data being realized through the integration of information, computer and communication technologies with traditional hardware and analytical thinking. It is going to affect the very nature of decision-making in crop management and every device and machine engaged in field production. Before elaborating on the “fusion,” I will briefly review the individual evolution of machines, sensors and models.
Since the beginning of agriculture, man has sought devices, such as tools, pumps and plows, to improve the efficiency of crop production while reducing labor and conserving resources. These devices were first operated by hand, later pulled by animals, and lastly powered by engines. The transformation of agriculture by machines in just the last 100 years has been truly amazing. As reported in a 2005 USDA bulletin entitled “The 20th Century Transformation of U.S. Agriculture and Farm Policy,” about 41% of the workforce — 22 million work animals and a few, newly invented, gasoline-powered tractors — were involved in agriculture at the start of the 20th century. By the start of the 21st century, slightly less than 2% of the workforce, 5 million tractors and a few work animals were active in agriculture. During this 100-year transformation, the number of farms in agriculture decreased by 63% while the average farm size increased by 67%.
A sensor is a device that converts a physical stimulus into an action or signal. Sensors have a history of development similar to machines. Beginning as simple devices that recorded a change in sound, motion, heat, pressure, light or other physical phenomena, sensors quickly evolved over the last 100 years into sophisticated arrays and networks. Sensors are ubiquitous in everyday life. They automatically open entrance doors in businesses, control lighting and heating in homes, detect the amount of fuel in cars and set off alarms in case of fire or gas leaks. Sensors can be placed locally or be remote, such as on aircraft or satellites.
Several recent advances have made sensors more applicable to agriculture. First, they have been coupled with radio communication. Today, a sensor placed in the field can measure some physical phenomenon, convert that measurement into an electronic signal and then transmit that signal with electromagnetic waves in the radio frequencies to a distant base station. This sensed measurement and its transmission by radiowaves can be done automatically, freeing up the need for someone to be on-site to retrieve recorded data.
The second advance in sensors is miniaturization. Sensors are gradually becoming smaller and smaller while still performing as their larger counterparts. Miniaturization is possible due to the use of new materials that require less volume, reduction in the size of electronic circuits and the exploitation of newly discovered physical, chemical and biological properties. Miniaturization, at or below the molecular scale, is called nanotechnology.
A third advance is the efficiency and cost reduction in the manufacturing of sensors. Sensors are becoming cheaper to make, which allows for more of them to be placed in the field at the same cost.
The fourth and last advance is the ability to combine sensors in networks. Sensor networks through their measurement and transmission of signals in spatial arrays over time can create a dynamic, two-dimensional and even a three-dimensional picture of some physical phenomenon.
A model is the mathematical representation of the physical world. Through parameters and equations, models mimic or “simulate” the properties and processes of some physical system. Models have existed on paper for more than 100 years, but their modern-day identity is linked to computers. Computers, through program code and machine instructions, can computationally execute the mathematical equations defining a model many times faster than a human can do by hand. Computer-based models can input and process data at mind-boggling rates. Furthermore, model-processed data or output can be presented in many visual forms, such as graphs and maps in support of management decision-making.
The Coming Fusion
With this background, it is easy to appreciate the fusion of machines, sensors and models. In the coming revolution, there will be a virtual “command” center running farm operations. Sensors flying on aircraft and satellites overhead in conjunction with those judiciously placed in fields and on tractors will measure physical, chemical and biological properties important to crop production. These sensor-based measurements will be converted to electronic signals and transmitted by radio to the command center. Base computers located in the center will receive the transmitted signals and deliver the data embodied in them via the Internet to models in the cloud. The models will process the data and pass back products in the form of tables, graphs and maps, depicting the state and changes in environmental and biological phenomena impacting crop development and growth. The same models will pass back recommendations on courses of action given status of the phenomena monitored in the field.
For example, a plant epidemiological model, inputting data collected in a field, may predict the incidence and severity of a disease important to crop yield. The model may recommend the timing and amount of a fungicide to minimize yield loss and control the spread of the disease. A farm manager would review the model-generated products and recommendation, and then choose a control tactic based on past experience and the available resources on hand.
If the choice is a fungicide as recommended by the model, a precision agriculture program could generate a variable-rate application map. This map would specify the rates of a chosen fungicide to be applied on a field according to the pattern of disease interpreted from sensor data. The variable-rate application map could be delivered wirelessly to spray equipment and, with GPS, guide the proper placement of the fungicide across a field. The fungicide application would change the progress of the disease, which would be indirectly monitored by sensor-recorded, environmental conditions. In a continuous cycle of sensed data, model processing of data and the incorporation of model products into precision agriculture programs, information would be generated to support management decision making during a growing season.
As precision agriculture evolves, it will play an important role in driving the demand for the fusion of machine, sensors and models. It will provide programs that allow a farm manager to act on model products. The same programs will support management decisions by guiding the operations of machines.
The new tech development reported in this issue represents small steps toward the realization of this fusion. With each new development, precision agriculture, along with machines, sensors and models, will increasingly provide decision-makers with information at an unprecedented scale and level of detail.