page1
The following position papers were received from participants at the remote sensing workshop.
· OPPORTUNITIES AND LIMITATIONS OF IMAGE-BASED REMOTE SENSING FOR
PRECISION AGRICULTURE............................................................................................................ 2
M. Susan Moran, Edward M. Barnes, USDA Agricultural Research Service
Recent advances in technology for variable rate material applications, with concurrent advances in global
positioning systems (GPS) and the ubiquitous use of geographic information systems (GIS), have provided a
powerful analysis tool for precision farming (PF). These advances have also led to intense informational
requirements. Image-based remote sensing may provide the timely, spatially distributed information on crop and
soil conditions that is needed to implement precision chemical and water applications and field operations.
· ACHIEVING RESPONSIVE AND EFFICIENT RESEARCH IN REMOTE SENSING
APPLICATIONS FOR AGRICULTURE ........................................................................................... 8
Thomas J. Gilding, American Crop Protection Association
The success of agriculture in the future will depend on how well performances in international markets, production
yields and efficiencies, natural resources conservation and environmental protection are optimized. Achieving
these greater levels of performance will demand more intensive management inputs throughout agriculture,
therefore, increasing requirements for more informed decision-making.
· FINAL REPORT OF THE AFBF GPS/GIS TASK FORCE............................................................. 12
Information Technology Advisory Committee, American Farm Bureau Federation
As recently as mid-1992, IBM market research indicated little or no market potential amongst farmers
for information systems using remote sensing data. That all changed in two short years. Suddenly, star
wars technology originally developed for the military became commercially available. Computers got
smaller, faster, more tolerant of harsh agricultural production environments and prices came down
dramatically. All of this brought some leading edge farmers to the conclusion that the time was right for
another major evolution in agricultural technology. The technology, they found, could help them
manage inputs and crop production on many smaller areas of each field, rather than treating each field
as a single homogenous unit.
· REMOTE SENSING REQUIREMENTS FOR AGRICULTURE .................................................... 16
George Seielstad, University of North Dakota, Upper Midwest Aerospace Consortium
Primary Requirement: Decision Support Information. [Paper in Outline Format]
Position Papers from the Workshop on
Remote Sensing in Agriculture in the 21
st
Century
October 23
rd
-25
th
, 1996


page 2
OPPORTUNITIES AND LIMITATIONS OF IMAGE-BASED
REMOTE SENSING FOR PRECISION AGRICULTURE
M. Susan Moran Edward M. Barnes
Physical Scientist Agricultural Engineer
moran-AT-tucson.ars.ag.gov ebarnes-AT-uswcl.ars.ag.gov
USDA-ARS U.S. Water Conservation Laboratory
www.uswcl.ars.ag.gov
4331 E. Broadway Rd, Phoenix, AZ 85719
Recent advances in technology for variable rate material applications, with concurrent advances in
global positioning systems (GPS) and the ubiquitous use of geographic information systems (GIS),
have provided a powerful analysis tool for precision farming (PF). These advances have also led to
intense informational requirements. Image-based remote sensing may provide the timely, spatially
distributed information on crop and soil conditions that is needed to implement precision
chemical and water applications and field operations.
Information requirements for PF fall into four general categories:
• real-time information for on-the-go management,
• information on seasonally-stable conditions (mapping long-term variability),
• information on seasonally-variable conditions (mapping short-term variability), and
• information required to determine cause of variability and develop a management strategy.
Conventional means for acquiring such information rely on tractor-mounted instruments, within
field point sampling using conventional instrumentation, crop yield monitors, soil and crop
models, expert systems and decision support systems (DSS). Image-based remote sensing
approaches could be used to supplement or supplant some of these conventional measurements,
and in some cases, provide information that is unavailable using conventional means. There are at
least eight shortcomings in current information gathering methods that could be remedied using
image-based remote sensing approaches:
1. Conventional univariate kriging is inadequate for converting point samples to field maps.
2. Combine-mounted yield monitors have limitations in resolution, accuracy, and flexibility.
3. Currently available soil maps are not suited for most applications in PF.
4. There are few viable means for monitoring seasonally-variable soil/crop characteristics.
5. Methods are needed for determining the cause of variability.
6. Spatially-distributed information on meteorological/climate conditions is needed.
7. Available digital elevation data are too coarse for within-field management.
8. Due to lack of timely information, time-critical PF applications are not being addressed.
Image-based remote sensing has potential to address each of these eight shortcomings, either
directly or in combination with other measurements or models, using common wavelength
regions at spatial resolutions of 1 km or less: reflected radiance in the visible, NIR and shortwave
infrared (SWIR) wavelengths (0.4 - 2.6 mm), emitted radiance (3-16 mm), and backscatter of
synthetic aperture radar (0.9 to 25 cm referred to as SAR).
One of the greatest obstacles to incorporation of RS images in PF will be the inherent limitations of
currently-available sensors. Satellite-based sensors have the advantages of good geometric and

page 3
3
radiometric integrity; the disadvantages include fixed spectral bands that may be inappropriate
for a given application, spatial resolutions too coarse for within-field analysis, inadequate repeat
coverage for intensive agricultural management, and long time periods between image acquisition
and delivery to the user. Though sensors aboard airplanes, helicopters and zeppelins will be able
to meet the requirements for fine spatial resolution, flexible and narrow spectral bands, frequent
repeat coverage and quick turn-around times, the previously-discussed difficulties in image
registration may preclude such data from many applications.
We propose that the following rules could be used to evaluate the proper sensor pixel resolution,
image delivery time, and repeat cycle relative to the size of the management unit, the
requirements for turn-around time, and the required revisit period (as defined below).
1. The relation between sensor pixel resolution (PR) and the PF management unit (MU) is
a function of the sensor signal-to-noise ratio (fS/N ) and the geometric registration accuracy
(fRA ), where
PR = MU/(fS/N +fRA )
1
.(1)
2. Turn-around (TT ) time is the total time the user can afford to postpone treatment while
waiting for the desired, processed information; it must be greater than the sum of the
image delivery time (TD ) and image processing time (TP ), where
TT > TD + TP
2
.(2)
3. Revisit period (RP) is the user’s requirement for repeat image acquisitions for the
specific farm management application; for sensors on a fixed repeat cycle (RC), it should
be a function of the potential for cloud interference (fC ) and for scheduling conflicts with
other users (fS ), where
RC = RP/(fC +fS )
3
.(3)
Additionally, one must consider that the total cost of instrumentation and processing cannot
exceed the perceived economic benefit.
Based on Eqs. (1)-(3) and estimates of M[J size, minimum turn-around time, and required revisit
period, an assessment could be made of current aircraft- and satellite-based sensors and upcoming
satellite-based sensors (Table 1). The following conclusions can be made:
· Qualitative images (without sensor calibration or signal atmospheric correction) from aircraft-based
sensors will be useful for limited, but important, applications, e.g., converting on-site
samples to field maps, mapping soil/ crop "anomalies", providing a quick assessment of crop
damage.
· Quantitative images from aircraft-based sensors will be quite useful for PF, particularly in
monitoring seasonally-variable soil/crop conditions and determining the cause of the
1
Due to adjacency effects with optical sensors and speckle with SAR sensors, fS/N ª10 for both
optical and SAR sensors; fRA can be 1 for most registered satellite-based images and 10-20 for many
aircraft-based systems with automated or semi-automated registration procedures.
2
For many satellite-based sensors, TD is generally no better than 24 hours and such "rush"
products are very expensive; best estimates of TP for images from both satellite- and aircraft-based
sensors is 16-24 hours.
3
Most studies report that for optical RS, fC ª4 (depending on location); for SAR RS, fC ª1. For
nonpointable satellite-based sensors, fS =l; for pointable sensors, fS ª4.

page4

4
variability. Rapid advancements in digital camera technology should improve the utility of
aircraft-based systems.
· Current satellite-based sensors have little potential for PF (coarse spatial/temporal resolution).
· The upcoming commercial satellite-based sensors will be suitable for many PF applications.
For example, the Earlybird satellites (planned launch 1997) will provide RC=3 days, TD =15
min, and PR=3 m panchromatic and 15 m multispectral (visible/NIR). The Quickbird satellites
(planned launch 1998) will improve the RC to 1 day and PR=4 m for multispectral data. None
of the upcoming commercial satellite systems will support thermal or SAR sensors.
· The advanced, high-resolution remote sensing systems aboard DOE, DOD and NASA aircraft
should be considered for mapping "seasonally-stable" conditions such as soil type (during
fallow periods) and yield variability (toward the end of the growing season).

page5
5
Table 1. Evaluation of RS as a source of information for PF Applications using sensors aboard small aircrafts
(where Ar : raw image data and Ac : calibrated data converted to values of reflectance, temperature or
SAR backscatter), sensors aboard currently-orbiting satellites (CS), and sensors planned for future
commercial satellites (FS). The check mark (÷) indicates that the application is appropriate for the
designated sensor; ÷L indicates that the application is appropriate, however the fields must be large;
and ÷W indicates applications which are only appropriate “within fields” because the data are not
calibrated and cannot be reliably compared over time or space.
EVALUATION OF RS AS A SOURCE OF INFORMATION FOR PF
APPLICATIONS
Ar Ac CS FS
1. Converting point samples to field maps
1a On-site measurements of soil and crop properties could be combined with multispectral
imagery to produce accurate, timely maps of soil and crop characteristics for defining
precision management units.
÷ ÷ ÷L ÷
2. Mapping crop yield
2a Multispectral images obtained late in the crop growing season could be used to map
crop yields with approaches as simple as regression or in combination with agro-meteorological
models
÷ ÷ ÷L ÷
2b Remote sensing information could be combined with crop growth models to predict
final yield ÷ ÷L ÷
3. Mapping Soil Variability
3a Multispectral images obtained when soils are bare could be used to map soil types
relevant to PF with approaches based on models and/or on analysis of single or multiple
image acquisitions
÷ ÷ ÷L ÷
3b Maps of spectral variability (obtained under conditions of either bare soil or full crop
cover) may prove useful for revision of maps of management units
÷ ÷ ÷L ÷
4. Monitoring seasonally-variable soil and crop characteristics
4a Soil moisture content ÷
4b Crop phenologic stage ÷ ÷
4c Crop biomass and yield production ÷ ÷
4d Crop evaporation rate ÷
4e Crop nutrient deficiencies ÷W ÷ ÷
4f Crop disease ÷W ÷
4g Weed infestation ÷ ÷
4h Insect infestation ÷W ÷ ÷
5. Determining the cause of the variability
5a RS could provide accurate input information for agricultural decision support systems
(DSS) ÷ ÷
5b RS information could be combined with agro-meteorological models to determine cause
of soil/crop variability
÷ ÷
5c Hyperspectral sensors could be used to determine cause of soil and crop variability ÷ ÷
6. Mapping spatially-distributed information on meteorological/climate
conditions
6a Multispectral images of coarse spatial resolution and fine temporal resolution should be
used to produce local or regional maps of meteorological parameters such as insolation,
PAR, rainfall, and others
÷
7. Producing fine-resolution digital elevation data
7a Accurate, fine-resolution DEMs could be produced from stereopairs of aerial and
satellite images ÷
8. Addressing time-critical crop management (TCCM) applications
8a For TCCM, multispectral images from aircraft-sensors could be used as a quick means of
assessing the extent of the damage and identifying units for damage control ÷ ÷


6
Herein, a case has been made for the potential benefit of remote sensing for PF. The real challenge
is to develop a system that will be readily adopted by the agricultural community. Our
experiences have confirmed that the factors cited by other researchers for successful technology
transfer of any innovative agricultural program will also work for acceptance of RS in PF; that is,
• Early interaction with the producer is essential.
• The system must be based on the clients needs (identified by the client).
• A gradual implementation of new programs allows the user to maintain an understanding of
the new
• technology.
• Participants must understand the operation of the program.
• Providing information is not enough; users need help assembling the information and
applying it.
• Ownership of a system affects farmers' attitudes and behaviors.
An infrastructure that may have promise for incorporating aircraft- or satellite-based RS
technology into PF is illustrated in Figure 1. The four "entities" portrayed in Figure I illustrate the
four requirements for skills and knowledge necessary to produce the three intermediate products;
actually, a single company could encompass the skills of the first three entities and provide the
final product to producers. Until an infrastructure similar to that illustrated in Figure 1 is in place,
there is little hope for widespread adoption of image-based remote sensing for PF.
Future work should be focused on determining which RS applications listed in Table 1 are most
economically beneficial and technically feasible. Season-long experiments with ground-, aircraft-or
satellite-based sensors designed specifically to investigate the economic and scientific viability
of RS products for PF applications should be given high priority. These experiments should be
designed with input from the end user (farmers and consultants), and the potential commercial
provider. Such validation will provide the confidence in RS that is required for technology transfer
and eventual commercial development.

-------------------

7

7
Image Provider
Images of reflectance, temperature or
backscatter registered to farm coordinates
Remote Sensing Specialist
Images of crop/soil conditions, e.g.
weeds, crop stress, soil moisture
Crop Consultant
Maps of management units for
variable rate material applications
Producer
Figure 1. An infrastructure that may lead to widespread adoption of image-based remote sensing
for site-specific crop management.
Reference:
Moran, M.S., Y. Inoue and E.M. Barnes, Opportunities and Limitations for Image-Based Remote
Sensing in Precision Crop Management, Remote Sensing of Environment. 61:319-346. 1997.
Skills and Knowledge:
Engineering
Computer
Optics
Remote Sensing
Skills and Knowledge:
Physics
Agronomy
Remote Sensing
Modeling
Skills and Knowledge:
Modeling
Farm Mgmt.
DSS, GIS, VRT
Agronomy
Skills and Knowledge:
Farm Mgmt.
Var. Rate Tech.
Computer
GPS

-----------------------------

9

9
3. Evaluating information requirements of major agricultural decisions.
Identifying the spatial and temporal information requirements of major decisions in
agriculture provides a basis of need for research in remote sensing applications.
Knowing the technical parameters of the information requirements is the link for
finding specific remote sensing applications in agriculture. Grouping and prioritizing
information requirements according to, 1) their potential economic or environmental
values of the decisions supported, and 2) leveraging common information requirements
among different categories of decisions establishes the need and justification for remote
sensing applications and supporting research.
4. Developing multi-disciplinary approach, including decision makers involvement to research.
Inputs and coordination among multi-disciplines are critical to the success and
responsiveness of research in remote sensing applications. The kinds of disciplines to
include would depend on the nature of research to be conducted, but most often would
include the basic biological sciences in agriculture An essential condition for responsive
research is the input and feedback from decision makers at key stages of research
projects. Decision makers must be partners in research that supports their decision-making.
5. Maximizing utility of existing research or technology in planning research projects.
An important step in planning research is the search for existing research that would
complement the planned approach of a research project Research from nonagricultural
remote sensing applications should be an area given special attention. Efficient research
demands reasonable efforts maximizing utility of existing research. Ideally, new
research would be reserved for "filling the gaps" that remain after existing research
findings.
6. Establishing realistic research completion schedules.
Schedules for completing research projects should be based on realistic projections of
time for achieving the research objectives. Influence of budget cycles or time lines for
research publications should be kept to a minimum. The major consideration is that
completion schedules should not arbitrarily affect the quality of research. Efforts should
be made to keep factors that affect research project schedules, other than achieving the
research objectives in a timely and efficient manner, to a minimum.
7. Requiring accountability of completed research.
A final step in completing research should be research management accountability to the
funding source(s). A significant requirement in this accountability should be; an
assessment of the research results in terms of achieving the originally stated objectives,
how the research is being disseminated for further utility beyond the intended
agricultural application(s), and recommendations on future areas of research raised by
the research just completed.

---------------------

10

10
Agricultural Decisions Having
Spatial/Temporal Dimensions
AGRICULTURAL DECISION MAKERS
1. Agricultural Producers:
2. Input Suppliers:
Agricultural loan suppliers
Agricultural insurance (underwriters and brokers)
Crop consultants
Labor suppliers
Land owners/brokers
Farm equipment (manufacturers, distributors and wholesale)
Fertilizers and pesticides (manufacturers, distributors, wholesale, applicators)
Seed (producers, distributors and wholesale)
Energy suppliers
Veterinarians
3. Output Services:
Commodity trade
Feed processors
Food & fiber processors
Futures (brokers)
Livestock (buyers, distributors, meat packers/processors)
Storage
Transportation
TYPES OF DECISIONS
1. Assess Natural Resources Quality/Quantity
2. Determine Optimum Fertilizer and Pesticide Application
Rates/Timing/Locations
3. Determine Optimum Harvest Timing/Locations
4. Determine Optimum Planting Timing/Locations
5. Determine Crop Acreage
6. Identify/Assess Environmental Risks
7. Identify Conditions Affecting Crop Yields
8. Identify Existing/Forecast Future Levels of Pest Risks
9. Identify Locations with Desired Vegetation/Vegetative Qualities
10. Identify Locations with Optimum Crop Growing Conditions
11. Predict Crop Yields

-----------------

11

11
SPATIAL DATA INPUT PARAMETERS
1. Air Temperature (historical, real-time and forecast)
2. Crop moisture (real-time)
3. Crop reflectance (real-time)
4. Crop temperature (real-time)
5. Crop yield (historical, real-time and forecast)
6. Landscape features (characterize)
7. Leaf wetness (historical, real-time and forecast)
8. Precipitation (historical, real-time and forecast)
9. Radiation (historical, real-time and forecast)
10. Relative humidity (historical, real-time and forecast)
11. Soil moisture (historical, real-time and forecast)
12. Soil nutrients (characterize and forecast)
13. Soil organic content (characterize)
14. Soil temperature (historical, real-time and forecast)
15. Soil texture (characterize)
16. Soil water holding capacity (characterize)
17. Topography slope/elevation (characterize)
18. Vegetative profile (classify)
19. Wind speed/direction (historical, real-time and forecast)

---------------------------

12
12
FINAL REPORT
OF THE
AFBF GPS/GIS TASK FORCE
Information Technology Advisory Committee
American Farm Bureau Federation
www.fb.com
Background
As recently as mid-1992, IBM market research indicated little or no market potential amongst
farmers for information systems using remote sensing data. That all changed in two short years.
Suddenly, Star Wars technology originally developed for the military became commercially
available. Computers got smaller, faster, more tolerant of harsh agricultural production
environments and prices came down dramatically. All of this brought some leading edge farmers
to the conclusion that the time was right for another major evolution in agricultural technology.
The technology, they found, could help them manage inputs and crop production on many
smaller areas of each field, rather than treating each field as a single homogenous unit.
In early 1994, three Iowa farmers, Varel Bailey, Tom Dorr and Lon Crosby were referred to the
American Farm Bureau Federation (AFBF) by the Iowa Farm Bureau Federation. They shared with
AFBF officials their vision of Global Positioning Systems (GPS) and Geographic Information
Systems (GIS) as being the future of agriculture. GPS/GIS was discussed at the AFBF’s Board
Meeting in June.
In July of 1994, AFBF President, Dean Kleckner, named a task force of state Farm Bureau
presidents. Its purpose was to "address the needs of producer decision makers in terms of data
ownership computer hardware and software needs, and the necessary Farm Bureau
organizational and financial structure if any, needed to deal with the GPS/GIS issue." The task
force was charged with bringing its recommendations back to the President in time for
consideration at the December, 1994, meeting of the AFBF Board of Directors.
Task Force Methodology
At its initial meeting August 25, 1994, the Task Force divided itself into three subcommittees and
drew up an initial list of topics for which internal white papers were developed.
The Task Force met a second time on October 12 and used two of its subcommittees to consider
the white papers and make preliminary recommendations to the financial subcommittee. This
meeting ended without a recommendation for the financial subcommittee.
At this point it was obvious that the whole issue was very complex and was moving so fast, on so
many fronts, that President Kleckner extended the Task Force’s life to March, 1995.
The Task Force then met again on November 17. Because of the urgency of a number of issues, the
Task Force produced an interim report with seven recommendations Eat it believed the AFBF
Board needed to act upon at its December board meeting.

--------------

13

13
In the interim, Purdue University was contracted to produce an interdisciplinary report looking at
as many of the GPS/GIS information systems issues as possible, strictly from a farmer perspective.
The Task Force met for the fourth and final time on March 2-3, 1995, to develop its final report and
recommendations.
Findings of the AFBF GPS/GIS Task Force
The Task Force finds that critical mass has been reached in the potential to use Global Positioning
Systems (GPS) and Geographic Information Systems (GIS) in farm-level decision making and
management of agricultural production systems. The key elements are now in place to allow most
farmers to begin to adopt some of this technology within the next ten years. These elements
include:
1. reliable yield monitors have been perfected,
2. the military GPS locational signals are now available for civilian use allowing latitude and
longitude to be determined with the 4 to 6 inches accuracy needed for farm equipment while it
is moving through the field,
3. personal computers are now big enough and fast enough to handle the gigabytes of
information that can potentially be produced from each farm field,
4. communications systems using fiber optics, cellular and satellite transmissions have greatly
expanded capability to transmit information and data, and
5. agribusiness at all levels, ad a number of companies in the aerospace and electronics industry
are very interested in the technology and are gearing up to offer integrated sensors,
controllers, hay logic and systems research as well as a wide variety of other related products
and services to support agriculture’s use of this technological advancement.
· The Task Force also finds that farmers arc buying yield monitors as fast as they can be
produced. Farmers are asking cooperatives and consultants to help interpret the yield maps
and other fertility information.
· The Task Force finds that the technology of gathering information is far ahead of the
understanding of how to use it to help farmers make decisions. Knowledge-based decision
making systems need to be developed quickly in order to allow this technology to fulfill its
promise.
· The Task Force finds that university research into GPS/GIS and precision farming systems
has, for the most part, been piecemeal, not interdisciplinary, and not coordinated.
· The Task Force believes that on-the-go sensors for soil, plant, and growing conditions are
crucial to fully developing the concept of precision measurement of many field conditions and
ultimately the precision placement of farm inputs. The Task Force finds that relatively few of
these types of sensors are readily available, (yield monitors, an organic matter sensor, a nitrate
sensor and some limited machine vision sensors for detecting weeds are the ones the task force
is aware of.)
· The Task Force finds that databases are the basic building blocks of this knowledge driven
system. The Task Force believes that control of access to a database holds the key to control of
14

14
that segment of an industry. Any bias in access to databases will create "haves" and "have-nots"
within production and marketing agriculture.
· The Task Force finds that standards for electronic transfer of data inputs and outputs are
lacking and that standards for physical connections of one piece of equipment to another are
also lacking. These need to be resolved as quickly as possible through appropriate national
and international standards efforts without stifling innovation.
· The Task Force finds that an "open architecture for electronics and data compatibility is the
preferred method for development of this information system, as opposed to many
proprietary architectures that would impede transfer of information, Open architecture would
allow competition in the marketplace over time to produce better solutions to problems.
· The Task Force finds that access to high capacity communications infrastructure is critical to
the development and use of this technology, especially in rural communities
· The Task Force finds that many special interests adverse to agriculture see this technology as a
key to furthering their policy agendas.
· The Task Force finds numerous questions exist regarding the issues of intellectual property
rights, the ownership of farmer generated data and the public policy implications of this data
These issues need to be resolved as-quickly as possible to protect the interests of producers in
this emerging technology.
· The Task Force finds tremendous interest in GPS/GIS and precision farming systems among
Farm Bureau members, as evidenced by the large turn out at special conferences held at the
AFBF Annual Meeting in St. Louis.
· Finally, the Task Force finds there is no coordinated overall GPS/GIS development effort.
There seems to be some interest within industry for surfacing a coordinating entity for such an
effort.
Recommendations of the AFBF GPS/GIS Task Force
General Recommendations
1. The Task Force recommends that AFBF work to insure farmer ownership and control of
farmer generated data and of databases compiled from such data, and to make sure that
systems research generated from the databases provides the products and services which
farmers need to move agriculture forward in a knowledge driven, competitive, international
marketplace. The number one priority of this effort should be to secure legal protection for
intellectual property rights of farmer owned GPS/GIS data.
2. The Task Force recommends that AFBF further explore the idea of a non-profit entity as a
vehicle for protecting farmers’ precision farming data and for developing the necessary legal,
business and computer capabilities to fully develop profitable farm, and off-farm,
opportunities for farmers from this new technology.
3. The Task Force recommends that AFBF work to assure that farmers and rural areas have
access to high capacity communications systems.

----------------

15
External Contacts
4. The Task Force recommends that AFBF work with state Farm Bureaus to make sure that the
appropriate researchers at Land Grant Universities, Agricultural Research Stations, ARS, and
electronic and technical institutes arc apprised of the Task Force’s findings and that there are
major research needs in the areas of sensor technology and knowledge-based systems.
5. The Task Force recommends that AFBF continue to work with the Ag Electronics Association
(AEA) as long as the relationship is beneficial for Farm Bureau and the AEA moves fast
enough to position farmers at the forefront of GPS/GIS opportunities.

---------------

16

16
REMOTE SENSING REQUIREMENTS FOR AGRICULTURE
George Seielstad
gseielst-AT-aero.und.edu
University of North Dakota
Upper Midwest Aerospace Consortium
September 26, 1996
1. Primary Requirement: decision support information.
1.1. Information, not data
1.2. Answers, not questions
1.3. Solutions, not problems
1.4. Conclusions
1.4.1. The only farmers and ranchers who will be interested in remote sensing
data are those who have the possibility to change their practices based on
the information. Every remote sensing product must answer the farmer’s
question, what would I do differently if I had it?
1.4.2. During the growing season, farmers have little time to manipulate data.
They need information presented concisely, enabling quick decisions for
how to act upon the information.
1.4.3. Applications of remote sensing should permit interactivity, so that
individual users can arrange the information into a form most relevant to
their needs.
1.4.4. All remote sensing images present historical information, namely the
conditions at the time the image was acquired. The most useful information
will incorporate forecasts or models, advising what conditions will become
and therefore what to do about them.
2. Reasons remote sensing has been used little in agriculture
2.1. Unfamiliarity. Farmers and ranchers not aware of what is available, how to
interpret it, how it can help.
2.2. Access to information has been inconvenient; the user has to initiate acquisition.
2.3. Cost-Benefit Analyses are scarce. How does remote sensing information increase
income and reduce expenses?
2.4. Cost. Images are too expensive, given that their usefulness has not been
demonstrated.
2.5. By themselves, devoid of context, remotely sensed images are not very valuable.
F. Conclusions
1. Education and training are essential for creating a demand for remotely
sensed products.
2. The market needs "pump priming." Until value is demonstrated and success
stories are generated, demand will be small and commercial products
unprofitable. This industry must be treated like the Internet or the GPS
system: government funding and no-cost data until the market is
established.
3. Products must be syntheses of various sources of information in addition to
remotely sensed images.

-----------------

17
a) Weather is the dominant factor in crop yield and quality. Farmers
should be encouraged to install weather stations distributed across
their fields. The integration of microclimate information with yield
maps and remotely sensed images will be more useful than any of
them individually.
b) In situ measurements made in the field with modern yield monitors
must be layered in GIS’s with remotely sensed data.
c) GPS-guided records of applications of water, fertilizer, pesticides,
herbicides, seeds must be incorporated into the same GIS systems.
d) Historical comparisons provide context. Examples are this week’s
vegetation index vs. last week’s, vegetation at this date vs. the same
date last year or vs. the average on this date of the last five years, etc.
e) Crop models must be refined so that the progress of a crop during a
growing season can be projected to its final yield.
f) Regional, national, and international comparisons are useful.
4. Information must be accessible locally. The information providers need to
be part of a distributed system, not a centralized one.
a) Agriculture practices vary with region. Generic products need to be
fine-tuned to local characteristics.
b) Individual farmers and ranchers prefer to interact with familiar
organizations of a size that is not intimidating.
c) The flow of information needs to be multi-directional. Providers and
users of remotely sensed products function interactively as co-equals.
The goal is to create "learning communities", in which
experiences are shared, questions asked and answered, ideas for
new products discussed.
III. Changes that make Agriculture ready for remote sensing
A. The Global Positioning System now assigns precise place and time to every action
and observation.
B. Yield maps display variability across individual fields, triggering a demand for
explanations.
C. Yield maps are acquired at harvest time and therefore can only be used to change
practices during the next growing season. Remote sensing could provide potential
yield information during a growing season in time to take corrective actions before
harvest.
D. The Information Superhighway makes distribution of information quicker and
easier. Individuals can directly access information, not needing hierarchies of
providers trickling it down to them.
E. The new Freedom to Farm Agriculture bill thrusts decision-making onto the
individual businessperson. Wise decisions must be based upon accurate and
current information. The demand for such information will increase dramatically.
F. More data from satellites is going to be acquired than ever before. EOS and
commercial satellites will provide a rich menu of spatial and temporal resolutions,
as well as spectral responses. Cost will still be a factor but perhaps competition will
keep it reasonable.
IV. Special needs of agriculture
A. Frequent information--i.e., at least once per week or 10 days--during growing
season

-------------------

18
B. Fresh information, i.e., 48 hours old at maximum, preferably