Don Nielsen started the discussion: Suppose I am research leader of 10 people. I have $1 million per year to spend on all agricultural problems before I retire in 10 years. Numerical techniques will become more powerful. Small scale will become larger. Data will be mined. We will not go into the field, but we will go into the field in the literature.

Logsdon: What will we do about soil structure?

Clothier: Should we teach students at universities the Boltzmann transform?

van Genuchten: No, do not teach this; we need geostatistics, instrumentation, GPR [ground penetrating radar]. What is needed is information on preferential flow and non-equilibrium flow. S o we need to know soil structure.

Nielsen then asked the speakers to do two things:

1. Ask one question that needed to be answered (assuming they had the money), and 2. Give a provocative statement.

Rodriguez-Iturbe: We need to know N mineralization as a function of time. We need pdfs [probability density functions] of N as a function of climate.

McBratney: I want raw data of nitrate in real time.

Rodriguez-Iturbe: Numbers won't be repeated in the future. I want relations, models of N.

Gardner: Dennis Keeney, M.B. Kirkham, and I studied this and got parameters. But we couldn't carry the research on, because of no money. [See: M.B. Kirkham, D.R. Keeney, and W.R. Gardner. 1974. Uptake of water and labelled nitrate at different depths in the root zone of potato plants grown on a sandy soil. Agro-Ecosystems 1:31-44.]

Jardine: How can we reduce the uncertainty of data? What is the role of temperature in models?

Prunty: If we consider temperature, we will improve models--e.g., include viscosity.

Clothier: Transport processes blow away small scale chemistry.

Warrick: What kind of model will give us the best results? We need criteria to decide what model is best.

Clothier: We need to ask questions. What is the definition of plant available water? What is the water-retention curve? An important question to ask is, "How deep is the root zone?" We need measurement techniques. Mesocosm and field experiments will give the best answers in soil physics. It is futile to do experiments to validate complex mechanistic models. We can use models to transcend scale. Time domain reflectometry (TDR) for soil physicists is the best thing since bubbles in beer. Learning things at different scales and validating mechanistic models are important. We need to say, "Here's what I measured and here's what I predicted." We need better models for root water uptake.

van Genuchten: We can waste time validating models.

Clothier: I design an experiment and then I apply models to it. The objective is not modelling. We get new understanding of data from the model.

Goovaerts: If I can go to Australia on a sabbatical leave for a year, what should I do? Develop new tools? My provocative statement is: In measuring random variables in a field, what do I model?

Rodriguez-Iturbe: Do we judge a model by the question it poses? Would any statistical model do?

Loague: I don't have to change jobs to come here. I have been in Hawaii and at Berkeley and now Stanford. My question is: In the future, will we spend more money (effort) on legacies or more effort on rigorous labelling regulations? My answer is the latter: We should not chase legacies. Assessments are being made to dig up Central Valley and take it to Nevada. Legacies are being created now.

Clothier: The best use of models is to design the next experiment. What should we measure?

Pachepsky: My question: What is soil texture doing along slopes? Distribution of heavy clays and rare events--how do they affect soil erosion? Soil erosion modelling tells us how things move down slopes.

van Genuchten: Motivation for this research is that 50% of arable land is on hillslopes. We need to know the transfer functions.

Warrick: The scale-dependence depends on the model.

Or: My question: What is the role of biology in soil physics? "I am not a soil physicist," people say, as if it's a contagious disease. Soil physicists don't take into account biology. My provocative comment is: Discipline boundaries need to be broken down. We need to reach a diverse audience. Soil physicists tend to look into themselves. They need to look out--e.g., interact with the petroleum engineers. Most of the things I do I have borrowed from other disciplines.

Ewing: Biology creates biofilms on surfaces. If we get organic deposition, we do not get perfect wetting. Microbial debris makes aggregates stick together (polysaccharides). Soil physicists ignore biology, because it is too difficult. We need to get soil mechanics into soil physics.

Boersma: What is the role of soil physics in biology?

Milne: My question: To what extent is a free energy view used routinely? My provocative statement: I am very glad to be here. My wife is a professor of biology. She won't let me return home without mentioning lack of women at the conference. Ecological groups now have to have one third of the speakers be women to get support.

Ewing: One woman ecologist that we invited to speak turned us down!

Or: Free energy is basic.

Pachepsky: I taught at Moscow State University and there were 90% girls in the class. This hasn't changed much.

Babovic: Marc Parlange needs to be careful in overselling remote sensing. Thirty years ago it fell flat. We need to be careful about dimensional analysis. We need to be cautious about the way we organize data.

McBratney: Any reason?

Parlange: We are here at Iowa State University. Do we really know how to predict groundwater recharge better than Kirkham and students did in the 1960s? Philip and de Vries did work in the 1960s; there is Wilford Gardner's classic paper in 1960. [See: Gardner, W.R. 1960. Dynamic aspects of water availability to plants. Soil Sci. 89:63-73.] Have we done better? No. Are there limits to predictability? Are we lacking new ideas? Separately, I would like to pick up on Brent Clothier's comment. He leaves the ground and goes to the tree. Fluid mechanistics of the atmosphere are no different than soil physics. We need to clarify the role of temperature.

Clothier: Sap moves through plants and soil physicists don't study it. It is rare for them to use Penman-Monteith. None of us learned groundwater recharge from Kirkham.

Gardner: I publically disavow the 1960 paper. My "Irrigation Science" paper is my disavowal. [See: Gardner, W.R. 1991. Modeling water uptake by roots. Irrig. Sci. 12: 109-114.] We CAN do better. We spent $6 billion on Yucca Mountain and what did we learn? Marc's question has not been answered concerning Don Kirkham.

Wendroth: My statement: The next important work in field scale processes will be a better estimation of ET (evapotranspiration). We need better attempts to predict water conditions. I now ask the Kirkham students here: What has emerged that is new? What would have helped you 30 years ago?

Clothier: I disagree. I run an annual water balance model. For example, on Day 65 I can assign a value for ET to it. I can use the model for different crops. For all 27 years, predictions are pretty much the same. What really matters is rainfall. It is highly variable in time and space. It has a great impact on the water balance. At the other end, what matters in measuring is the rain-intercepted loss. How much rain gets to the soil surface and how fast does it reach the surface?

Logsdon: Landscape and terrain determine ET in Iowa.

Klock: Doppler effect is a huge improvement in models.

Parlange: Interpreting radar reflectometry can be hard. I agree with Brent and also Ole. The shape of the leaf is important.

van Genuchten: We need to know root water uptake as a function of water stress and soil stress. This is largely unresolved and we struggle with it.

van Schilfgaarde: We have spent too much money on lysimeters and measuring ET. There is a better way to spend money.

Corey: A tractor now costs $100,000. Back then (35 years ago), a tractor cost $25,000 and corn was $3.00 a bushel. Corn is still $3.00 a bushel. We now have computers and TDR--developments in 35 years. Computers help. In 1962-1963 we had neutron probes. They became important. We did not make major strides. Once we drained soil, we have done little. Crop scientists have made major advances in genetics (e.g., Bt crops). We are using new codes. We are not making major strides. We are not doing things that we are capable of doing.

Prunty: We have studied real situations. We have kept restrictions from being worse than they might have been.

van der Ploeg: Developments in agronomy are gradual. Major advances are made when people come into the field from another area, like Don Kirkham, a physicist, did when he entered soil physics. Darcy and Liebig were similar. One of the purposes of this conference was to include people outside of soil physics, and we included the hydrologists.

McBratney: [A comment to follow-up on Milne's comment: Alex said that 60% of the people doing bachelors of soil science at his university in Australia are women.] My question: Will we get a new theory, like a fractal theory? Can we write equations for different scales? My provocative statement: The density of soil physicists is not enough for the future.

Klock: I would like to answer Jack Corey. We can use basic soil physics to solve societal problems. Jack has done this. I have done this. I am a private consultant, and my clients pay for soil physics advice. There are three parts to modelling: First predict; second, exhibit; third, explain after the fact.

Corey: I have heard lots of interesting things here at this conference that are very important.

Ewing: My question: There are millions of gallons of spill--e.g., at Hanford, WA. Don Kirkham worked on continuum procedures, but there is spatial variability. How do we extend models to 2 km? My provocative statement: discrete models have a lot to offer us that continuum models can't do.

Prunty: Agriculture is concerned with a thin soil layer.

Lewin-Koh: My question: How good are the assumptions in our models?

Loague: Models work in the calibrated range and, outside it, they probably do not work. Concerning the leaking tank at Hanford: The pH is 12. What is going to happen to cesium and plutonium in the tank? This is a real problem that exists today.

van Genuchten: Modellers focus on the model. We have to look at the problem.

van Schilfgaarde: A model can be anything, not just a computer model.

Nielsen: This is the end of the discussion. Thank you all for taking part.

Boersma: The conference is over. I can say it was a successful conference. Rienk van der Ploeg will be summarizing the conference. [This summary appeared in "CSANews," February, 2001, p. 12.]

---