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

| Ag 21 Agenda |

SESSION 1:  WHO ARE THE END USERS AND WHAT INFORMATION DO THEY REALLY WANT (AND ARE WILLING TO PAY FOR)?  Moderated by John LeBoeuf, Fordel, Inc.

SESSION 2:  WHAT REMOTE SENSING CAPABILITIES WILL BE AVAILABLE FOR AGRICULTURAL RESEARCH AND COMMERCIAL APPLICATIONS IN THE NEXT TEN YEARS?  Moderated by Jack Estes, UC Santa Barbara

SESSION 3&4:  HOW WILL THE DATA BE ANALYZED AND WHAT ARE THE PRIORITY SCIENCE ISSUES FOR THE NEXT DECADE?  Moderated by Larry Biehl, Purdue University and Susan Moran, USDA - ARS

SESSION 5&6:  HOW CAN AGRICULTURAL AND/OR ECOLOGICAL MODELS BE COMBINED WITH REMOTE SENSING DATASETS TO DEVELOP NEW SUSTAINABLE AGRICULTURE OR OTHER ENVIRONMENTAL RESOURCE MONITORING PROGRAMS?  Moderated by Edward Sheffner, NASA Ames Research Center and Mingua Zhang, UC Davis

SESSION 7:  WHAT ARE POTENTIAL IMPEDIMENTS TO TECHNOLOGY TRANSFER TO THE AGRICULTURAL/RURAL SECTOR?  Moderated by Wayne Mooneyhan

SWOT:  S=Strengths of Remote Sensing
The end users need to know the strengths of remote sensing.

Stress detection
It needs to be identified so end-users know what to look for, whether it's water (over and under-irrigation), areas in fertility, disease, weeds, nematodes, varietal differences, etc.  It's time to show where this can be shown in precision agriculture, which includes forestry.

Change detection
This is very important for producers, if you can detect change, whether it's change you're hoping for or a change that needs to be looked at and remedied.

High quality road maps for field boundaries
If this is a way to get a lot of users to look at remote sensing, then it's time to push this to the forefront.  Even if it's a single use -- it's a way to get a lot of users to look at remote sensing.

Time management
This is important to the producers and the people out in the field.  They need to know where to spend their time in the problem areas.  They can put their highest quality people into trying to solving these problem areas.  You can put lower level field scouts into other areas.  Time management:  where to put your resources that will make the biggest impact.  Another time management thought was that more acreage may be able to be covered by a single consultant.  This has not been verified yet.

Match up Sensors, resolution, scale, crops, GIS, GPS, etc. with Specific Crops
We don't have time to discuss each one, but if a sensor works for a crop or a certain part of detection, it needs to be identified so users know where to go.  They need to know what resolution will work for certain cropping patterns.  We're still in the beginning phase of trying to learn this from users.

SWOT:  W=Weakness of Remote Sensing
Weaknesses need to be identified to the end users.  If there is a weakness, bring it forward so they know.
1. There are no or few spectral signatures that diagnose problems.  Ground-truthing is necessary.  Even with forestry, it's still important to have people in the field to detect what's going on at the ground level.  We cannot at this time, sit back [and identify a problem or disease from an image].

2. Sensors must be matched up properly with products, as users will remember and talk about failures of technology.

3. It is not possible to predict all future uses of remote sensing.  Let's look at taking a few of the resources and promoting them off the strengths and let the users decide what remote sensing can do.

SWOT:  O=Opportunities for Remote Sensing
1. On-farm research.  Farmers and consultants want research to move away from small plots and be brought on to the farm where they can observe progress under field conditions. The growers are willing to participate.  We need to look at this as an opportunity and get involved.  It's time for action.

2. Interdisciplinary teams need to get involved in R&D.  We're talking about agricultural science, environmental science, biological science, ag economics, crops, soils, irrigation, the ecology, etc.

3. Outreach to the existing agriculture professionals through continuing education requirements, professional societies, NASA, ag extension and specialists.  Advisors in the state of CA have to have 20 units (on average) of continuing education every year to keep their license in the industry.  Next week CAPCA (California Agricultural Production Consultants Association) is having their annual meeting that includes 1,000 professionals and consultants.  We need to reach these people and bring them into awareness of remote sensing.  Also NAICC (The National Alliance of Independent Crop Consultants) have meetings all across the U.S., CCA (Certified Crop Advisers) have professionals here.

4. Professional societies could provide guidance and training and education needs.  If we have a need for curricula to be modified, let's look to using the professional societies to provide guidance.  They (American Society of Agronomy, Crop Science and Soil Science)can bring us into focus for training needs for existing professionals and also the educational needs of the new people coming into this industry.

5. NASA could - and should - demonstrate technologies to growers and consultants to help overcome low awareness. Bring the demonstration out and show the technology [for short-term and long-term management].  People don't know what airborne sensors can do or what thermal imaging can do.  Are there videos that can be made to show them to the people?  Can you use the Internet and try to make a web site? Are there grants that can be developed?  Utilize CAPCA, NAICC,  and CCA.

6. Agricultural extension and specialists could provide unbiased information.  I have heard that people look to ag extension and specialists to make sure that [the technology] will remain scale neutral.  It is very important that the small producer will be able to access information.

7. Quality attributes can be connected to quantity.  This will help producers.  It will become very important to control quality of commodities and not just focus on the quantity harvested.

8. Enhancement of crop models and decision support systems.  Any technology that can assist in the decision making process will be welcome.  People talk about information overload -- they can't be boggled down by extensive models that they have to run.  They're looking for the bottom line:  what do I need to do now, when, what's it going to take?

9. Yield monitors developed for commodities that are machine harvested in a single pass will bring in new users of remote sensing.  There are people wanting yield monitors.  The monitors will show variability leading people to ask what's causing the variability.  This will then lead individuals into looking  for help from remote sensing technologies.

10. Improvements in variable rate technology will also increase the demand for remote sensing.  The related issues that improve variable rate technology may increase the area threshold for farmer action, which I currently hear is 25% of the field.  A lot of times, unless 25% of the crop is threatened, the farmer won't jump on a problem unless it's a high value cash crop or drip irrigation, etc.  We may see improvements that may lower the threshold.

11. Use existing commercial sensors and systems to promote satellite and aircraft platforms.  We  - the end users- don't even know what can be used by farm managers utilizing technology already operating.  Demonstrations of technology applications with agricultural consultants are needed.

12. Space Grant:  education, research and extension.  If it's feasible, embrace it with open arms -- we need all the help we can get.  Make use of these potential opportunities.

SWOT:  T=Threats to Remote Sensing
Questioning of fairness and liability by users.  1)  Can remote sensing data purchased by a user be accessed by the competition?  Can a grower, shipper or supplier access someone else's information? 2)  Can remote sensing data be used against a grower for damage to the environment?  This is a very hot topic.  They perceive Big Brother wanting to see what's in their backyard.  3)  Can remote sensing data be used against a consultant who made a recommendation?  Are they liable in a court of law?  If you're expecting the consultants to be the bridge between farming and science and technology, that question is going to be asked.

Findings
1. The users of the raw data may be a small subset of the total users.  I thought more people would be looking at the digital information.  This was going to drive the users to bring it in more so than color infrared photography that we had 20 years ago.  What I was thinking was that raw data extracted out of digital information could be used to help look at crop growth curves.

2. Ensure product reliability.  I don't have the answers as to how to address these issues, but we're finding there is a need to produce imagery with no false positives.  Identifying a problem that isn't actually there in the field points to unreliable products which have clutter.

3. The culture of the growers is not a problem.  We heard earlier that culture was a problem.  We need to decide if it's a problem or not.

4. Age is a factor  -- I see this at the ground level in agriculture.  There are some growers who would be willing to retire before they have to bring in all these technologies, and I see the younger generations coming out of college embracing the technologies.  The father, patriarch of the operation, is limiting that use.

Final Thoughts
It is time to make the agronomists happy.  Try to learn something new every day.  Let's not pretend we know it all.  The more we learn - the more we realize that there's more  - to learn.  Search out golf courses located in agricultural production areas and get out and golf. [Laughter]  Dr. David Zilberman brought this up yesterday.  We may not realize how entwined golf is with agriculture production.  For every consultant or adviser out pounding the pavement, there's someone else out on the golf course putting away, talking about making big trades of agrichemicals and supplies.  Even as we're getting into the global markets, we have produce buyers, produce sellers that are out there.  There are deals being made on the golf course.  [Laughter]  Let's not think like Michael Jordan.  Let's be like Tiger Woods.  Let's get out there and think we can do it.  Let's get out there and do it.  This is America. It can be done.  Be a tiger, don't be timid.  You must "want" to serve this industry or the ag industry will beat you up.  You must be pro-active, not reactive.  Any questions?  Thank you for your attention.

P:  [Inaudible] JB:  There are groups that are looking at individual property rights(IPR).  We have some standard groups working on IPR.  One brief comment about the Farm Bureau -- there is a task force trying to reach growers and find out what they're looking for in protection of individual property rights .

Remote Sensing in Agricultural Litigation
P:  There is no question that remotely sensed data has been used in agricultural litigation across this country and will continue to be used in agricultural litigation.  There's nothing you can do about that.  JB:  Remote sensing for determination of  crop damage is already being  used, but I think that the bigger picture is on the environment.  Hypoxia was mentioned in one of the sessions as an example.  P:  Is it only crop damages?  It's agricultural pesticides, runoff, overirrigating that causes landslides, etc.  It's going to continue.  It's going to be used by county, state, Dept. of Transportation, stuff that [farmers] have no control over.
 You say the culture of the growers is not a problem.  I think the culture of the growers is a problem.  JB:  The older farmer may perceive this as a threat and may put him into retirement a little quicker.  P:  From a commercial standpoint, they're not threats to remote sensing, they're marketing opportunities [Laughter]  JB:  That's a true entrepreneur!

Satellite Platforms and Sensor Capabilities
With respect to satellites, Bill Stoney gave us an excellent talk in the morning.  Everyone agreed that the satellites were pretty well covered.  There were some additions -- intermediate resolution systems and our participants repeatedly wanted to get down to particular kinds of application.  What are you going to use this sensor for?  Remote sensing in agriculture can go anywhere from stress detection at the inner field level up to global models in weather and climate.  So we’ve got scalar function all the way from tremendously high resolution to tremendously large areas covered.  The additions included some satellites and capabilities like EOS AM-1, MODIS, LIGHTSARs, GOES, NOAA, SACKAJAWEA (thermal infrared).

Aircraft Platforms and Sensor Capabilities
With respect to aircraft data, there are pretty much aerial platforms and sensors to do anything we want to do in the area of agriculture and remote sensing.  We can fly UV sensors, visible multiband, short wavelength IR, middle IR, thermal IR, K-band, X-band, L-band, P-band, C-band, UHF radars, passive microwave devices, hyperspectral sensors, multi-depression angle, multi-polarization, active systems, sounders, LIDARs, laser-induced fluorescence. We can use aircraft from single engine, light aircraft flying low altitudes, to high altitude, high performance aircraft and multi engine planes.  Temporal and spatial resolutions are also covered.  If you want to fly the Centuri Romano near Padua in Italy, you may need something that’s got less than one meter resolution.  Average field sizes here are some 2.5 ha.  Then you have the mid-west or San Joaquin Large Field agriculture and some of the very large field of the Kazakh Republic.

Areas Where Research is Needed
[Our discussion was based on] the bias of the people in the room.
1. SAR, multi-frequency, multiple-compression, multi-polarization is an important area.
 
2. Thermal infrared.  We need to look at the sensitivity tradeoffs.  Very fine temperature resolution, very fine emissivity.  Resolution is very expensive technology because of the cooling that has to go on.  Maybe they can go to uncooled detectors and thermal technology will be a lot cheaper to fly.
 
3. Calibration.  Bill Stoney says that all sensors are calibrated.  Yes they are, but we talked about inner-sensor calibrations and multi-sensor multiple platforms calibrations of sensor -- how to do that in a more effective way.
 
4. Atmospheric correction.  We’re not taking the atmosphere out of the signal.
 
5. Soil moisture.  This is one of the reasons that SAR came up.  People said soil moisture is really important for agriculture.  We need surface soil moisture, seed zone soil moisture, root zone soil moisture.  It needs to be looked at from satellite, aircraft and potentially even ground to give us better data.
 
6. Sensors for the chemical makeup of crop residue.  How much organic carbon is left in the field.
 
7. Ground-truth sensors.  This can be difficult to do and extensive.  More particular applications ... ground truth with the aircraft and satellite data.
 
8. Through-put.  Getting the information out to the farmer within a reasonable amount of time. George Seielstad, up in University of North Dakota, has formed alliances with people to get the data out to the farmers in near real time.  The key is real-enough time -- real enough to be able to do something with the information you get.
 
9. Hyperspectral.  We need more work on hyperspectral.  Some people agreed that more work needed to be done, others disagreed.

Conclusions and Recommendations
1. The fundamental conclusion:  At this time, it appears that ag remote sensing capabilities will exist and be available over the next ten years to support a wide variety of research and commercial applications for agriculture.
 
2. Research is still needed in a number of areas.  But for more of the focused research that needs to be done, people want to use remote sensing commercial applications.

There were no specific recommendations from our people.  The recommendation I would include is that  NASA should fund research in some of the areas identified.  Anyone in our group have other comments?

P: ... The problem of getting data, as far as satellites are concerned, the data is available ...  What we have to do to solve the problem of throughput.  Certainly NASA should... recognize this need to get data to users quickly.  P:  There is a major study that the National Academy of Science is conducting on international copyright as part of the world information policy organization.  All three of the academies wrote a letter to the Secretary of Congress that said Congress had proposed a law that said Europeans had the right to copyright meteorological data.  This would fundamentally change the science return in the U.S.  If you get the head of the Institute for Medicine, the head of the National Academy of Engineers and the head of the National Academy of Science all signing a letter to the Secretary of Congress.  That’s a pretty strong statement.  Maybe this could make the whole science community come together to protect our access to science data.

P:  The research in the key areas you identified that were mostly sensor technology.  How do you see the balance?  Should NASA only develop sensor technology?  Or does NASA have a role in the private projects... ?  JE:  I don’t know quite how to respond to that for the group.  The way I would respond to it is that I think NASA has enough technology up there.  If NASA is going to do anything in their technology development program, it would help agriculture if they put money into SAR and into thermal infrared.  A lot of people think that the other technologies are far enough ahead that we need to learn how to use those better.  R:  I would just make the comment that I don’t think it’s perceived that the capability to do agriculture is up there already.  There needs to be more effort in the implementation of applications, and communications to the folks that are going to implement those applications -- what the capabilities are.  For instance, if I think I might have a use for remote sensing, how do I find out where to get this information and who to work with?  JE:  I agree with you, but that’s not a question that this group has to deal with. I’ve been conducting agricultural research with remotely sensed data since 1964.  I think some of the research we did then might be of interest to TRW today.  I talked to them and they said they went back through the literature on agriculture and went through the papers.  I know for a fact that there’s stuff out there.  It was done at Purdue, Kansas, Michigan, Berkeley, etc.   I’m an academic.  How do we say, “Here are some experts in agriculture and remote sensing that companies can come to and ask our opinions?”  That’s not what I do.  That is not what is happening yet.  In many cases a lot of knowledge from the 70’s and 60’s is being forgotten.

P:  Just a quick story.  The AgRISTARS program, LACIE, produced a huge archive at the Johnson Space Center (JFC) of literature.  The program ended.  UC Santa Barbara offered to take that archive and put in our map data library and make it available and accessible to everybody.  Cost estimates at $20,000 were given and we’d put it all on line.  NASA Goddard took it, got it, and it sat in boxes.   Two years later, they sent it to UC Santa Barbara because nobody at Goddard used it.  It’s there, the whole archive.  They’re not on-line, no monies were given with the data.  That’s just an example of how much great literature is out there.  I know where it’s sitting, but I don’t have the time to go through and catalogue all of it.

P:  I just want to make sure that people are aware (I think most are) that when you’re talking about sensor development.  On a regular basis ... there are and will be opportunities such as the coming ESSP proposal call which is announcement for the community at large.  NASA had a very laid-back attitude and we just specified the kinds of things we wanted to get out of the data. It was us, the industries, universities, PI’s who put together themes to address the problems.  What I’m getting at is the agricultural community can advance together and develop effective ways to get a major infusion of funding if it is done right.  A good way to monitor these things is on the Mission to Planet Earth homepage (www@hq.NASA.gov/office/mtpe/).  There’s an area there for research announcements.  R:  I’m in agricultural remote sensing research.  It seems like NASA decided not to do agricultural related remote sensing, because they were basically asked not to.  Maybe because of lack of interest on the overall agricultural community.  We have people trained in agronomy at my branch with the idea is that there is interaction with NASA on agricultural problems.  That interest or stated interest would help quite a bit. JE:  The thing I just want to tell you is that there are a lot of great people in NASA who know a heck of a lot about agriculture.  Just like when they shifted from using Landsat and went to AVHRR, there was a good reason for that.  People will do what they’re paid to do if it works.  I was doing a lot of agricultural research.  I personally am very anxious to see NASA go back and go back in a way that coordinates and cooperates with, collaborates with USDA and other Federal Agencies in an appropriate fashion.

We broke down the results into different areas:  science issues, the need for pilot projects, remote sensing and ancillary data, levels of products, issues related to data analysis with resolution, spatial, spectral, temporal, understanding several items under image characteristics.  We had some other weather opportunities related with communications, the transfer of data, the transfer of education, new sky[?] type ideas, longer range things that need to be done, some discussion relative to standards.

The Need for Pilot Projects
This came up several times during the three hours.  One thing that was talked about is the science for pilot projects.  What we have is a specific management question and working around building the projects around that.  One person recommended that they not be complex, but the individual question of a group of growers might have.  Recommendation:  The pilot projects should not necessarily be based on the research going down, but look at the question the grower has and work back upward to solve that.  Keeping track of how much is saved.  Trying to identify problems in the remote sensing technology at the lower level.  We try to work with the grower directly to find out what the problems are -- think in terms of the grower’s demands.
 
Remotely Sensed Data is not the Sole Product
You don’t necessarily want to think of a remote sensing image as just something by itself.  Provide it with other management information.  Also, it can be used as inputs into other models (canopy architecture, crop growth, spectral).
 
Yield Monitors and Predicting Yield
In particular, one of the big questions is how to use the data [for predicting yields].  If you’re looking at yields, now can you tell me in June or July? That will be [a question] at the local level, but also at the national and global level.  The data could be used with crop calendars or analysis done in the late 70’s early 80’s projecting that you’re taking remote sensing data across time.  The other comment brought in was mid-season predictions.

Determining yield in below ground crops.  A comment brought up by TRW is that a lot of times they’ve had problems relating biomass to production.  Biomass above the ground doesn’t relate directly to production below ground.
 
Variety of Data Needs
There is a need for the raw data and for highly processed data and reports.  There’s intermediate needs as well.

One comment I’d like to make is one doesn’t want to make a lot of generalities.  [For example], what resolution is required?  It depends.  How’s the data going to be used?  It depends on the area, the crop, the size of the farm.  It’s going to be hard to say that we’ve got one type of data product that a lot of people are going to use.

The economics of it -- At one time, it may not be profitable enough.  But if some foreign policy changes or the market changes, we’re going to have to be ready to [respond].  Something that may not have been economically viable before, [may suddenly become necessary or profitable].
 
1. The need for research and application of temporal data analysis.  In fact, some statements were made that [one person would] rather see money taken out of the hyperspectral or hyperspatial and just get data more often.  That relates to the crop calendar.  The issues that relate to that are cost of more data, but also knowing when to do it (which depends on the crop calendar) and inputting it into your cropping models.
 
2. Spatial resolution and moray patterns.  The issue that came up is that if your spatial resolution gets down to where the crop management practices and the same order as what your crops are planted, you come up with a moray pattern.  You’ve seen the effect of this in some data.  You want to think of it as the spatial resolution required as a function of the crop management process -- it depends on the crop and the use of it.
 
3. Image characteristics of understanding.  Trying to get the education out to the potential user or the buyer of data.  Part of the discussion was on calibration and defining the range of the sensors.  [However,] depending on what your analysis is, there are differences in what calibrations are required.  If you’re going to store data across time, then you need more detail.  There’s some information related to calibration and atmospheric correction.  You don’t want to do that on just a blind type basis.  We had students look at a flat piece of land that was fairly small in size, and when the atmospheric correction was done, it made it worse.  The variations due to the atmosphere was only a few percent, but the correction is only valid for 10%.  We added noise to it.  That’s something the user needs to be aware of.  The sun view and sun angle effects -- if you get the temporal data two days apart, there’s going to be a lot of variation.  There’s also the effects when you’re registering data -- multi-temporal, different projection systems.  Recognize that what you do with registration may affect your analysis.  If you start averaging pixels together and use second-order statistics, the data for the analysis are destroyed or become much less important.
 
4. Communication technology -- where does that stand related to the users and research?  It’s not going to be constant across the country.
 
5. Standards - specific instruments from companies that could be used together.  A grower involved in vineyards asked about a standard on projection.  He’s just getting involved in it, and he’s trying to get his database together.  He’s got this layer with GIS from one company, and a projection from TM, what’s best for him?  This is part of the educational process to recognize what you need to work on at the end-user level.
 
6. The SAR was mentioned as a definite need [in the long term] for agriculture.

[Question about bistatic] LB:  In bistatic that was one where you actually have your source coming from this angle and you pick it up at another angle, parallel.

Recommendations
1. A partnership within the government (with NASA or USDA, etc.) and their role is the technology transfer to the pilot projects.
 
2. A need for an agency to assist the end-users [assess and understand] the technology.  Whether it is through extension, [users] need somewhere they can turn to for answers and education.  If they’re working with the commercial people, they may want to have an “unbiased” place to go to try to reinforce what the commercial people are advocating.

P:  Three years ago, I went to work for NASA 50% and 50% for US Department of Interior.  I got shot at from both sides.  I came to realize that you had two different cultures in those two federal organizations.  On one hand, NASA has a lot more money for R&D than USGS does.  It’s an R&D agency, USGS is an operational agency.  USGS had a lot of people, but not a lot of money, and they never realized that NASA has a limited amount of time to do anything.  The people at NASA are overstressed.  When they would make a decision that the people in USGS didn’t agree with, USGS would internalize it into their institutional framework.  In other words, they couldn’t see how could NASA make that decision.  I would say that the guy at NASA typically has 1 1/2 hour to make a decision; they’re an R&D agency.  They won’t put somebody on it for 6 months to study it to death, and then make a decision that’s more reasonable.  It’s going to be the same way when we try to work with both NASA and USDA.  There’s been historical institutional differences between those two agencies.  All of us have to be aware of that.  I would also say that academics have to be sensitive to commercial interests.  Commercial interest have to be ...  You have to begin to understand that we’re working with a lot of diverse groups that all have their own cultures.  We have to work the best we can to reduce the levels of tension.  R:  In my business, we don’t have many resources available.  We had to work with all of the different groups, and I don’t think there’s enough recognition on what each group places an emphasis on.  If you look at the operations and see how each company... It really needs to be approached on a ...  basis on what you have to offer.  It takes a lot of time.

P:  I’d like to comment on the hyperspectral data processing.  ... tremendous work from sensor level all the way down to the ... distribution.  Namely you have to do the processing calibration and the fluorescence.  The most advantages to hyperspectral are seeing the signatures of spectral.  Other things, at this present time, [are done] manually.  There’s tremendous man-power hours, and currently what we’re working on is to automate the process -- automatically radiometrically calibrate and then classification.  At that point, you set up a spectral library that you can pull the signatures from the library and do classification automatically.  After that you store data and archive it.  Users can browse samples and then get the data from the record from archives without operators.  R:  How do you do the crop signatures that change over ...  How do you set it up?  R:  First of all, you need classify crops against trees.  Those big families we’re try to set up the  spectral.  After that, the next step is over the time period, what’s the signatures of the spectral [as it] changes.  It’s a long process.

P:  In our  group, we had some discussion of hyperspectral, and the general feeling of the group was there is a lot more that needs to be done with that... My research of hyperspectral is exciting, but I cringe when I hear people says “signatures.”  Because you have so many variables that are thrown into the background... R:  If you really want to do it, you need to set up experiments in the Southern San Joaquin valley where you don’t get ... depends on what you want to do with it.  [Decide what you want to do] -- identify the crop type, phonology, etc.  You begin to develop “signatures” every time you put that ... looking for variation.  Anybody ever think we’re ever going to get ... are totally wrong.  But you get close.  There are some commercial purposes -- what you want to do is get close and then let people make the final decisions.

P:  With Landsat we look at the field and identify crop types or [identify] the development stage.  But when you look at the hyperspectral in the field, there’s a lot of variation.  I see precision agriculture being able to take your ground-truth samples with every pixel of your hyperspectral.  Now we’re getting to the point where we can get some ground truth data that begins to support it. [Inaudible response] Did your group comment on the need for more frequent or ... existing kinds of technology?  LB:  I thought we had one person comment on multi-temporal.  That was the way ... [Inaudible responses followed by laughter]

[Editorial comments by Edward Sheffner]
ES:  Speaking about spectral signatures, about a decade ago we were working with the USDA on the competitor’s software.  It was software that USDA started work on in the 1970’s to do crop estimates using Landsat data.  It was tighter than a black box.  They had their Landsat data, they had their code, process it and come up with their area estimates, and they would never look at the imagery or the classifications.  We took a scene at random that included part of the Mississippi River.  They had the classification wrong.  That scene they had the entire Mississippi River as soybeans.  [Laughter] There was no problem with the code.  When they went back and looked at their training fields, it so happened that the training fields in that scene for soybeans they used that year were flooded [Laughter]  Of course they never looked at the imagery so they had no idea what was going on.
 I’d like to mention that the LANDSAT program is not an operational program.  Even if you just consider one satellite, it’s not an operational program.  There’s no backup.
 This question about the temporal resolution issue [is one that] I’ve been hearing for years.  There are many uses out there, when you ... more often than once every 16 days.  The only way you’re going to get that now at that resolution is to look at the inoperability[?] sensing.  Try to combine LANDSAT with SPOT with IRS with other similar systems.  You may not be able to for some applications, because of calibration problems but you may for others.

Issues Related to Sustainable Agriculture and Resource Monitoring
A lot of comments were about problems people were having with remote sensing, a lot of recrimination[?] since this is California.  A lot of the issues haven’t changed.  We did identify some issues, most of which have been repeated this morning.

Sub-Field Issues
In terms of within-field scale, the resolution we’re talking about is one meter.
1. Evapotranspiration.  Percent canopy cover of the field.
2. Crop stress especially looking at thermal imaging.
3. Frost warning prediction.  Apparently, there’s a problem with getting weather data for frost warning.  Time to take preventive action.
4. Soil characteristics, especially soil moisture.  There was a general consensus that except for soil moisture, it would be very difficult to use remote sensing to determine soil characteristics.
5. Within season time points.  These would be indicators of the vegetation in the field.  Where that field presides in its crop growing cycle and using that information to calibrate production models and potentially yield models.
6. Accurate geo-registered base maps for use in GIS.
7. Spectral resolution.  There was a limited discussion on spectral resolution for within field work.  It had to be good enough/suitable to be get a plant nutrient chemistry analysis.

Regional/Global Issues
The resolution scale is about 30+ meters.  It could be larger in some instances depending on the application.
1. Water pollution and water distribution.
2. Food commodity production.
3. Soils information including soil moisture...
4. Canopy plant structure for forestry and range lands... [inaudible comments]

Recommendations
In talking about the appropriate roles for the U.S. government and the commercial sector, there was some agreement that [the government] should take on regional and global issues, primarily.  The commercial sector could handle the subfield data needs.

There was a noted lack of city... input for agriculture issues.  There is a lot of ... available at the city ... level.  ... taking nature out of the Central Valley in California --urbanization in the valley and how that’s affecting ag lands.  There is some information coming in that we need use to do studies on ag/urban issues.

... accessed information.  This relates to issues brought up earlier on making sure that everyone who wants access to the data will have access to the data.  There’s no restrictions on ...  That’s all I have to report.  Anyone want to add anything.

[Inaudible comment] ES: How did they arrive at one meter?  Someone suggested it and no one disagreed or said it was too fine.

ES:  In the end, I think they want all scales.  There is a report ... done by the Forest Service.  They used the vegetation index, NDVI ... to predict fire potential.  We’re looking at two states, Idaho and Montana.  One year in 1993, the other in 1994.   All they’ve done is take NDVI measurements in spring, early summer and late summer, and fall in those two years.   They average six years of data, and these are actually plots of now minus the average of six years.  So in 1993, these huge green areas indicated that there was lots of fire potential and lots of vegetation --150% of average.  In 1993, late in the season we had 50%.  We’re talking about a farmer who might need one pixel, half a pixel or three pixels of this.  This is mostly dry-land farming.  He’s got a chance of making 50% of average or 150% of average.  We wonder if we’re helping him by advising him meter by meter crop systems. I think he’s overwhelmed by a regional pattern that goes way outside his boundaries.  I keep wondering if the resolution is right.  This one sobers me [since these farmers are] driven by factors over which they have no control.  R:  But that’s non-irrigated.  R:  Yes, but that’s where I am.  My own impression is that in 1994, you lost money and maybe you can lose a little less -- I can help, but maybe the advise for a precision farmer in 1993 is no more chemicals, fertilizers, etc.

P:  I’d like to make a comment about spectral and spatial resolution.  It’s going to be nation or past[?] dependent.  If you talk about trees, you might need one meter resolution but if you look at the areas and environmental impact by certain incidents[?], you don’t cover a lot of areas at 10 meter or 20 meters in impact regional changes.  In talking about the resolution of the spectral, I know that basically you have a ... indicators of stress as well.  You don’t have to use the thermal or visual infrared to detect vegetation stress... The resolution needs to be selectable.  In certain regions you need a finer resolution, but in other areas, like with infrared, you need a much coarser resolution because of the application.

ES:  Back to the question about how they came up with one meter.  There actually was a rationale for it that had to do with site specific farming.  The [?] on the back of the tractor was about 3 meters across, so you want to go to 1 meter pixels, so you can be sure of getting a pixel within that area.  P:  Also, in an orchard, you can’t identify ... so you don’t want to miss that. R:  ...Precision farming can be anywhere from looking at the region to determine what to do ...  So you really need to determine what type of measurement unit it is that you’re really looking at... People come to me and ask what the best resolution is.  I think what we need to tell people that ... it’s a function -- a function takes into account, not only the management with it, but also the signals ... ratio, sensor registration accuracy, products -- I think there are several things that need to be equated.  I don’t think it’s a difficult function to the client, nor is it a difficult function to solve.  I don’t think it’s a terribly nebulous question anymore, it’s quite easy to answer.

P: ... dependent on the specific issues that you’re addressing, what you’re modeling is, whether it’s a commodity, diagronomic[?] or field crop, etc.  It can be issues within each project.  It can be different issues depending on the application.  I suspect that ultimately, when we finally do develop restrictions so the growers can use remote sensing, they’ll be in relatively simple terms that the grower can interpret.  We’ll talk about ways that they should plan it.  We’re working for specific indices that the growers will be able to take out and use.  R: ...If we don’t get the scale right for that grower ... with having to control a specific space scale and time scale, they’re soon going to be frustrated...  When you do apply a treatment on the landscape at that appropriate scale, there will be an effect that the grower then sees and that can be measured through a yield monitor and other feedback mechanisms.  So the answer is we don’t know.  There are methodologies available to determine what that space scale is... [Inaudible comments]

P:  The commercial user is going to be the developer ... determines what resolution his data... No matter what we say or what we think, he’s got to be the one to requires one meter, 10 meters, 1/2 meter.  He’s going to be the one to design it.  You can’t [force him to take] 10 meters whether he likes it or not... R:  One of the problems.  We didn’t need one meter, multi-spectral data, but the other kind of information that you might want to go with it might be ... soil mass[?] ... Even if you have absolutely fantastic remote sensing data at that scale every two days, then you have a ... capability to get it out the door.  The other components they might need to get isn’t effectively in place.  R:  The question had nothing to do with that.  The question was remote sensing databases and their applications/models for the purpose of monitoring and environmental monitoring. [Inaudible responses] R:  ...he can’t say the farmer want’s this or needs that.  The needs are different.  The need is something you absolutely must have...  He wants to use remote sensing because he wants to do better.  But there’s nothing that I can do, as a commercial supplier of data or NASA or the universities can do, that can make him need remote sensing.  That’s the challenge to those of us who are trying to sell a service.

P:  I would like to throw the discussion up to issues that you think none of the groups addressed, but really are issues you’d like to have considered in the recommendations to NASA.  R:  ...the need for ... funding from NASA for education at the graduate and undergraduate levels.  It would be useful to identify some ... customer awareness.  Is it appropriate for this group to make the recommendation of key points that NASA ... We’d like to see some of these issues ... some are modeling, some are user awareness, etc.  Is there some program we’re trying to identify... I just wish they’d look at ... appropriate, prioritization.

P:  I thought the standards ... we need to discuss.  Based on the variety of the user, you cannot set up a standard and format all the way down to the level.  In between, you can set up, the government supplies the level onto some median level.  Then you need to tailor, pair it down to the individual users.

P:  I’d like to go back to John’s question about what efforts does this group see as appropriate for awareness, outreach, education activities on all levels?  How important is this component to the overall success of the technology?  R:  I work in the Napa Valley.  There’s transfer technology companies coming by, working with us all the time.  One way we try to identify who’s really serious about it -- we’re willing to offer the land, but they have to provide the product free of charge.  We offer all the personnel -- we expect the companies to, at the same time, have some input.  Now we’ve got these 35 weather stations that are owned by growers.  The company donated the equipment before anyone was willing to adopt it, and it spread all over.  We now have growers that are making contracts with companies, because they are aware of how valuable it is.  One of the things NASA or someone could do is take a product like Susan’s where you’re having a group of growers -- you need a base of people of 10-30 people to understand it, want to use it, and the provider will come in and use ... we have to have those connected across the country in different agricultural zones responding to each other and also make a contract with the provider.  If they take one photo and share it will 30 people, it makes the cost lower for everyone involved.  I don’t think the growers don’t want to develop the market themselves, but I do think that we can if we can get a base of enough growers in an area.

R:  I would like to point out that Bob Reconado[?] representing ... with Alex... Bob Harris are developing a U.S. program... work on these issues. [Inaudible response] meeting ... to specifically discuss long term pilot programs with NASA and USDA for the purpose of applying NASA’s  Mission to Planet Earth technology.  We’re going to take this report and hope to ... address awareness... distribute it to agricultural associations... [Inaudible response]  R:  There is a program that’s funded by the Educational Office in NASA.  It’s called The Remote Sensing Core Curriculum development projects or application projects in there.  We have requested to provide education modules and agriculture etc.  If you are interested in providing teaching modules from undergraduate, graduate and beyond, that’s one step that we are already working on.  [Inaudible responses]

P:  I just wanted to say (Jack touched on this) ... information program specifically directed to agriculture...  Eventually transition to the ASTRS.  Other opportunity that develop a monitor specifically directed towards agriculture.  Development of educational ideas-- that’s communicated well. [Inaudible responses]

P:  One comment I’d like to bring up is we talked about standards in the group.  One must be careful... Recognizing that they have to be connected to... R:  What you can say is if we want to work on standards, we work on the harmonization of the variety of standards of the different formats.  Data products that can be integrated into ... Harmonization is not standards. [Inaudible responses]

P:  ..Space Grant could be put together right now... [Inaudible responses] R:  We’re thinking about setting up a meeting at NASA headquarters with our Space Grant and how we can better utilize the Space Grant people that are in the universities in remote sensing. [Inaudible responses]

P:  ...It’s difficult to implement something national... They try to find ways in their states with their ... especially with Land Grants institutions and also Space Grant.  That’s where the work can begin.  For example, my institution... environmental institute which harbors both researchers, teachers and cooperative extension people working together in the area of research, education and extension.  Let’s be examples of modeling for Land Grant institutions ...Space Grant connected with cooperative extension in remote sensing.

Four Potential Constraints
1. The culture of the users.  It was generally felt that the farmers who are not remote sensing people are not prepared to accept a brand new, high-tech approach to farming.  It would require changes that may come very slowly.
 
2. Lack of interaction between product developers and product users.  This is generally true in most of our new technologies.  The developers push long before the users pull them.
 
3. Lack of interaction between the end users and data providers.
 
4. The perceived unfairness among the users.  For example, the farmer’s themselves don’t want their information handled too early by the commodities people.

There are new technologies that might benefit the growers in particular and ag business in general that should be transferred to them.  There was a general consensus by the group that some potential impediments existed.  The purpose of the group was to establish what those impediments were and who was responsible for or has the opportunity to respond to those impediments.  The group listed those impediments, developed a list of actions/activities that it takes to address those impediments, and identified a sector that could best address those issues.  The list was not prioritized due to lack of time.

Issues of Technology Transfer to the Rural Sector
1. Rural access to high-speed data link will take some time.  It was kind of a “chicken and the egg” problem.  Whenever the demand is there, those capabilities will exist.  There was a lot of discussion about Direct TV links that could be used at the present time.
 
2. Hardware and software are more of an institutional problem than a technical problem.  This confirms the earlier decision not to talk about this issue.  The hardware and software will be developed if innovative systems are developed and demand is increased.
 
3. The cost of hardware and software of space data systems is rather high and fixed.  Somebody has to pay for the data to pay for the systems.
 
4. The culture of the growers is generally resistant to new technology that will cause significant change.
 
5. There has been little or no input into the product development from the users themselves.
 
6. There has been little effort to integrate data systems to support the needs of agriculture and agribusiness.

Major Impediments and Discussion
User Awareness
The major impediment of technology transfer to the rural sector is that most farmers do not know that remote sensing satellites exist.  An action should be taken, led by NASA and supported by remote sensing industries, to collect a significant number of success stories and document them so that they can be put into educational packages, the farm press, trade journals, etc. to get information into the hands of the farmers.  There was discussion about the number of farmers that read the farm press, a significant publication in the ag community.  Recommendation:  Some small funding effort out of NASA to go to some group (university researchers and other researchers) to organize and package remote sensing information for maximum exposure to the grower/end user.
 
The Cost of The Data
The group noted that the decreased amount of research data and applications with the commercialization of the LANDSAT satellite. Since the commercialization in 1982, Purdue University has purchased 12 scenes.  Prior to that, they had purchased 4000+. Not only research, but applications and development essentially stopped with that price increase.  Recommendation:  NASA should consider sponsoring an application program where 1) data is provided by NASA; and 2) funding is provided for application projects to restart more research and applications of the data.  NASA and remote sensing industries should consider providing free data to the universities and research institutes involved in applications development.  At this point, the moderator promised to draft a statement from this workshop to NASA concerning an effort to shift some moneys from pure research into applications to try to restart the application work aimed at the agriculture community.  It will be included in the conference publication.  One of the reasons for that statement is that increased volume will decrease cost.

The data supplier should consider pricing data on a unit area basis as opposed to a scene.  If a farmer only has 200 acres, why should he buy 185 kilometer scene?

The providers and the new high resolution data sources should consider providing simulated data free of charge to universities and research institutions in order that the software and algorithms necessary to process that data can begin development prior to the launch of the satellite.  A statement on pricing policies should be developed and presented as an output from this workshop.
 
Lack of trained personnel in the applications of remote sensing to precision agriculture
The number of people being requested in remote sensing from the agriculture industries are greater than those being produced by graduate and undergraduate schools.  That is partially due to lack of funding to support the programs that students would be involved in.  Recommendation:  NASA and USDA should shift some research money to application studies and development at universities and research institutions.
 
Coordinate activities
The effectiveness of the efforts by government, industry and academia could be increased by an order of magnitude if they could more closely coordinate their activities.  It appears from the discussions, that the interest of those groups tend to converge at the end-user level.  Recommendation:  NASA should facilitate and fund an alliance among government, remote sensing industry and universities which would meet twice a year to address the issues that are constraining the transfer of this technology to the end user.  A major objective would be try to deliver datasets to research and application development communities in order to reduce the lead time between data processing technology and the data.  The alliance should be interacting as much as possible with the end user community.
 
Lack of an Applications Development Program supported by Government Funding
This was perceived to be the greatest impediment to the transfer of technology.

P:  I have a comment about the alliance.  I think NASA should form an alliance with the commercial sector.  [muffled]  If we’re going to have alliances within this group,  we’re going to have to make that happen.  Everyone can recognize... we’re on a fast track.  On the scientific side, we going to get ready and find ...  There’s a role for USDA and commercial to do the research whereas space industry can get what they want and Resource 21 can get what they want without stepping onto proprietary rights.

R:  There has been an attempt to form that alliance, what Tom Gilding required us to do with Ag Information Alliance last April.  Before we start another alliance related to agriculture, we ought to look at what Tom has already put together.  Tom basically has a philosophy that we needed to have all the input from the different associations (communications, computer, remote sensing, etc.) working together and well as all the industry groups.  It was the Alliance for Agricultural Information. They’re working as partners.  David Mack (Belmont) was representing the ag electronics association.  They have been working closely with the standards involved in hardware/software as well as all of the types of things from industry, so that the user doesn’t end up with a whole bunch of data that they can’t analyze because they have the wrong software or hardware package.  There is communications between the Agriculture Electronics Association and the Alliance of Agricultural Information.  They have cross membership.

P:  One thing that they could do is start a list server in which we could all communicate.  R:  We have something like that and could set something up like a list server pretty easily.

P:  If there’s some way of working out opportunities or get the federal people to spend some time in the industry and industry spend some time with the feds, and the university spend some time with commercial people.  We ought to promote that, because I’ve found that we’re not doing a very good job of training our graduate students for remote sensing industry jobs.  There’s a mechanism for working that out.  R:  As an academic, I can listen to the recommendations.  ...When private industry hears that, they’ll think the academics will go out and develop the product and go out into a private firm.  You’ve got to develop policies that say, universities will not take on operational jobs for the commercial sector or in this government environment that you have today.  [Otherwise] you aren’t going to go anywhere.  It’s great for us to do research, but I want to train my students better...

P:  That might be a good idea for NASA and private industry to cooperate.  Back to data cost; again, you’re dealing with data at different scales.  If you look at the LANDSAT scenes, when LANDSAT 7 is launched, cost will go way down, by at least by a magnitude of four.  So LANDSAT 7, if that’s the kind of data you’re interested in, will be a lot cheaper, a lot more readily available and a lot more data.  If you look at the data field scale at 1-5 meters, there’s no chance that NASA can provide data at that scale?  That’s pretty much left to private industry.  You’re going to have to buy that data.  NASA might purchase some and distribute it strictly for research and applications development purposes.

P: It sounds like the recommendation is that NASA should initiate some demonstration or pilot projects, and then we should look to other agencies (particularly the USDA and the commercial sector) to bring in commercial firms.  The USDA should make contributions to this effort, rather than just have NASA do the entire effort.  Will the commercial firms come with data and studies with the farmers?

WM:  That’s the idea behind the alliance.  One more comment:  There weren’t many growers or producers in the group today, and our recommendation was there to include and involve them.  If we don’t, the recommendations are always going to fade.  I don’t know how we need to make that happen, but we need to closely address the end users.