Proceedings of the Workshop on Remote Sensing for Agriculture in the 21st Century October 23-25th, 1996    Introduction

| Ag Workshop |

Executive Summary

The suite of these information intensive technologies is sometimes collectively referred to as Precision Agriculture or Site Specific Agriculture.  Because of the ability to provide synoptic spatial sampling at frequent repeat intervals and relate measurements to physical and environmental processes, satellite sensing is a key component.  These new technologies have many potential applications spanning the breadth of the agricultural industry, at all scales of organization from farmer, to cooperative, and professional societies; from farm machinery vendors, fertilizer and chemical companies, insurance, regulators, and commodities traders; to agronomists, consultants, and farm advisors.  A wide range of agricultural practices could be affected including potential environmental benefits that could come from minimizing adverse impacts by reducing external inputs and greater use efficiency.

Eighty leading scientists, aerospace engineers, agronomists, consultants, and agricultural industry representatives were invited to participate in a workshop to explore these issues at the University of California Davis.  The group met for two and a half days in October 1996 to discuss the issues surrounding the successful adoption of remote sensing technologies in production agriculture.  Participants provided their visions for future agricultural applications of remote sensing technologies and the impediments to the transfer and adoption of these technologies.  This workshop report includes summaries of the six keystone presentation topics and synopsis of the group discussions.  The report also includes papers and presentation materials that were provided by the invited participants as additional background information and position papers on this subject.

The next decade promises to be an exciting period for the application of space based remote sensing for Earth observation.   Proposed launches of fifty or more commercial and civilian satellites have been announced, that measure virtually all parts of the electromagnetic spectrum that can be transmitted through the Earth’s atmosphere.  Furthermore, some of these satellites will be available at spatial resolutions down to one meter in scale and have frequent repeat cycles over the growing season.  Even today, the availability of spatially detailed satellite information about crop and soil condition is unprecedented.  The optimism and enthusiasm of the participants for the future of satellite based information is apparent in body of this report.  Nonetheless, the report also clearly identifies areas where essential links between the data providers and the consumers are presently not in place.  End users perceive that NASA does not understand agricultural problems and lacks the institutional connections to provide products that will create demand for the data.  To successfully create a lasting demand for agricultural satellite data products, these missing links must be forged.

The overarching question for farmers, farm consultants, regulatory agencies, and commodities markets is  how to turn this flood of raw data into new information that is effective for decision making and the analysis tools that increase market profitability and security.  The consumer sectors of the agricultural enterprise need to make satellite data providers aware of their information needs.  For the aerospace industry, the critical concern is to understand customer needs, both current and potential, and provide useful products in a timely and cost-effective manner.  For the agrochemical, equipment, and seed suppliers, the burning questions are how to provide goods and services that support spatially explicit management decisions that have been based on inputs from satellite information.

For such sweeping changes to occur in just a few years in an industry as diverse as American agriculture assumes that strong competitive economic forces exist to support and accelerate the adoption.  Certainly, not all industry sectors will adopt data from these satellite sensors equally or rapidly.  Furthermore, the adoption of the full suite of precision agriculture technologies requires a relatively sophisticated and computer literate workforce.  This demand may interact with other forces in rural demographics and electronic communications to reach full fruition.  The ability to utilize this information assumes the university and NASA/USDA labs accept a critical role in providing the research that will develop the agricultural decision-making tools.  Much of this focus needs to be on addressing multivariate statistical models and cross-disciplinary approaches to crop ecology.  The Land Grant Universities and Agricultural Extension also provide an essential role in training the next generation of agronomists, and support retraining and continuing education for people currently working in the agricultural industry.

The report presents the workshop results describing the following six topic areas:

• Summary of past and present satellites and a review of remote sensing capabilities.
• Land observation satellites expected to be launched in the next decade.
• Lessons learned from past remote sensing agricultural programs.
• Information requirements for agriculture.
• Integrated decision making for profitability and environmental protection.
• Technology transfer and resistance to adoption.

The workshop identified several areas of recommendations.  These recommendations and supporting materials are developed more fully in the body of the report.  Here we identify general recommendations for sensor characteristics and research needs for remote sensing and agriculture.
 

General recommendations were proposed that are needed to facilitate adoption of remote sensing:

1. Develop more extensive continuing education programs to train experts in the agricultural sector to use and understand remote sensing data.  Training could take various forms through agricultural extension, college continuing education courses, professional societies, and various public and commercial short courses.
2. Develop application specific software and remote sensing data products.  Most sensors and data products are aimed at broad market acceptance, but rapid acceptance will come from specialty products.
3. Copyright regulations and/or pricing need to be relaxed or re-assessed to permit multiple users to benefit from data purchases.  We encourage “broad base” encouragement by adopting a low-cost pricing structure and that will promote a mass customer audience.  This approach should further stimulate new applications and penetration into presently undefined niche markets.

Recommendations for sensor characteristics:

1. Narrowing sensor bandwidths will allow development of predictive models based on actual biophysical absorption features.  The use of multiband thermal sensors, middle infrared bands, and high-resolution optical sensors will increase accuracy of yield estimation and other crop characteristics.  The number of sensor channels needs to increase to provide improved crop models.  Some experts consider that the use of modern imaging spectrometry has an important role in agricultural applications.   Possibly more than 1000 bands are needed to fully exploit the information from the UV to thermal infrared electromagnetic region.
2. There is a need for calibrated surface reflectance satellite data products so that multi-observation dates and multi-sensor data can be made with greater assurance of predictive accuracy.  Data needs to be georectified and spatially registered to a common map base for multidate comparisons and facilitate comparisons with other information from precision agriculture sensors.  Spatial registration becomes increasingly of concern at the 1-5m pixels that are the projected resolutions of several commercial sensors that will be available for agriculture.  Use of predictive crop models requires greater accuracy of data calibrations to enhance reliability of interpretations.
3. Improved analysis software using automated information extraction procedures are needed to match the larger band numbers, calibration, multi-date, and crop model inputs.

The workshop identified several priorities for research:

1. On-farm multivariate research is needed to fully understand the interactions of multiple sources of variation.  The implementation of pilot studies involving growers would improve acceptance and shorten the adoption period by ensuring that data products realistically meet users needs.
2. Development of multidisciplinary approaches to agricultural R&D.  Agricultural science and related disciplines (e.g., soil science and hydrology) needs to be approached from an ecosystem perspective for sustainable management.
3. Higher level (e.g., level 3 or 4 data products) has greater potential value to agricultural end users than less highly processed data.   “Higher level” data products involves converting the raw data into specialty products (like mapping soil organic matter, or crop yields) and presumes that basic image processing steps like calibrating to surface reflectance and spatial registration are essential components of data products.  Because these data products (including specialty products) are more experimental than operational, a schema for fully validating remotely sensed data products needs to be developed.  Validation must include both theoretical, in situ, and empirical validation methods to be robust.
4. Soil moisture is an important agronomic variable and more research, especially using passive and active microwave is needed.  Thermal sensors may improve better understanding water availability in the active root zone.  Other soil properties (e.g., organic matter, nitrogen) are also needed and monitoring of these properties will require use of the full solar to radar spectrum.
5. Research using other sensor information e.g., multiple SAR frequencies, view angle dependency, polarization, and thermal infrared is needed to fully exploit the information content of these sensor characteristics.
6. Research on laser technologies to determine canopy height and other properties.
7. Appropriate spatial, spectral, and temporal resolution for optimal information extraction.
8. Development of transfer standards for easy exchange and data fusion processes.
9. Need for linked crop models that use remotely sensed drivers.
10. Need to predict yield properties when growth and yield are poorly correlated.  Develop below ground monitoring capability.  Develop early stress indicators.

Respond to Introduction