| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
USDA-ARS Soil and Water Manage. Unit, Dep. of Soil, Water, and Clim., Univ. of Minnesota, St. Paul, MN 55108
* Corresponding author (jbaker{at}soils.umn.edu).
Received for publication November 20, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
Abbreviations: CFC, chlorofluorocarbon
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
|
SURFACE–ATMOSPHERE EXCHANGE |
|---|
|
TOPABSTRACTINTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
However, the barriers to more accurate measurement of surface exchange processes may be more conceptual than instrumental. The fundamental validity of eddy covariance has always been unquestioned, in part because of its powerful theoretical simplicity, but perhaps more because there were no fast-response sensors capable of rigorously testing and applying it. Now that there are infrared gas analyzers for water vapor and CO2, and tunable diode laser spectroscopy systems for a range of other trace gases, a mountain of data is accumulating, and a disturbing amount of it must be discarded or at least marked as questionable. Most scientists take great pains to regularly check the accuracy of their analyzers; the suspicion is growing that instrument error is not the problem. Rather, it appears that the assumptions of eddy covariance, primarily those of stationarity and surface homogeneity, are so restrictive that they preclude the continuous measurement that is necessary for a time-integrated flux of extended duration with acceptable accuracy. At a long-term site over a Douglas fir [Pseudotsuga menziesii (Mirb.) Franco] forest in British Columbia, careful winnowing of a 50-d data set that initially contained more than 2500 half-hourly CO2 flux values left a residual set with approximately 600 acceptable numbers; more than three-fourths of the data were rejected, most for either insufficient stationarity or lack of turbulent mixing (Drewitt, 2002). Problems of this sort have a diurnal bias—nighttime measurements are afflicted more frequently than those in the daytime, potentially introducing a dangerous bias in the integration of long-term CO2 measurements.
Also vexing is the mounting evidence that even when the underpinning assumptions are in place, there is a tendency of eddy covariance to systematically underestimate turbulent fluxes. Comparisons of latent and sensible heat flux measurements with surface energy balance often show a failure on the part of eddy covariance to close the energy balance by 20 to 30% (Twine et al., 2000). Wagner-Riddle et al. (2000) have concluded that this is so systematic that they multiply their flux data by 1.3, even though they use the aerodynamic method rather than eddy covariance, because the stability corrections that they use were developed using eddy covariance data as ground truth.
Why does eddy covariance come up short? There are a number of potential culprits—in fact, nearly all of the possible sources of error lead to diminution of the covariance. These include the spatial separation of the wind and concentration sensors, for which correction algorithms have been reported by Laubach and McNaughton (1999); non-zero vertical velocity of dry air at the point of measurement, presumably caused by advection or terrain irregularities, for which corrections have been proposed by Paw U et al. (2000); and missing high-frequency portions of the covariance, for which corrections have been proposed by Massman (2000). A more difficult problem may be hiding in larger-scale, low-frequency motion (Baldocchi et al., 1996). In principle, this could be accounted for by increasing the averaging period, but in practice, this may seldom be possible without violating stationarity constraints.
A more general problem with micrometeorological techniques is the assumption that the underlying surface is a uniform source or sink for the scalar of interest. We know that this is rarely the case; the burgeoning activity in precision agriculture is an implicit acknowledgment that field-scale variation in net C exchange is a widespread phenomenon, and this is in managed ecosystems where every action, from fertilizer application to tillage to single-cultivar planting, is geared toward homogeneity. The spatial variation in natural ecosystems must be generally far greater. Of course, small-scale variability, over distances of a few meters, is of little concern since it gets mixed out before it reaches the height of measurement. Larger-scale spatial variation, of the order of tens to hundreds of meters, is the problem. The direction, shape, and extent of the flux footprint, or the area contributing to the measured covariance, change constantly as wind direction and atmospheric stability change. When this is superimposed on underlying variability in surface exchange, it injects noise in the flux measurements, even in the absence of any sources of error, for which it is difficult to imagine an appropriate filter.
A thorough reader of micrometeorology literature over the past decade or so might draw two conclusions. The first is that in the area of flux measurements, the ratio of methods papers to application papers is exceedingly high. The second is that, despite this, more methods research is needed. These seemingly contradictory statements reflect the fact that current methodology is still inadequate to solve many of the environmental problems to which it potentially could be applied.
Soil Water and Solute Fluxes
Just as flux measurement has been at the core of micrometeorology, so has it been with soil physics. And as frustrating as turbulent flux measurement has been, the situation below ground has been even more bleak. Widespread concern about groundwater contamination has spurred extensive research efforts on the impact of land use practices on water and solute flow through soil, yet the only cases in which water and solute fluxes through soil have been directly measured have been in closed systems—either fully enclosed drainage lysimeters or tile-drained fields with impermeable lower boundaries. In almost all other field studies, fluxes have not even been measured but rather inferred from discrete measurements of soil water content or matric potential, knowledge of upper-boundary conditions, guessing of lower-boundary conditions, approximations of transport coefficients, and a numerical model. This approach can work tolerably well in a macroscopic sense to at least produce a visually acceptable match between measured and simulated water content profiles, but there is little evidence that it is a reliable estimator of flux dynamics, particularly for solutes. Of course, the primary reason for the dearth of such evidence is the lack of available flux measurement tools.
A promising recent development is the equilibrium tension lysimeter described by Brye et al. (1999). It is a stainless steel box with a porous surface that is buried in the soil and connected to a vacuum source. The vacuum is adjusted to maintain it as closely as possible to the separately measured matric potential of the surrounding soil at the same depth so that the water flux will neither converge nor diverge at the lysimeter. Periodically, the lysimeter is pumped out. The measured quantity of water and solutes yields the mean flux of each over the time period dating back to the previous emptying. Subsequently, enhancements of the method have included automated rather than manual control of suction and continuous measurement of water depth in the lysimeter, which provides better temporal resolution of water flux. Lentz and Kincaid (2003) have also described a controlled-tension lysimeter for collecting leachate. Gee et al. (2002) have described a simpler device that has no control of suction but is easier to emplace and operate. Such a device will probably work reasonably well in sandy soils, but flux divergence will likely be a problem in finer-textured soils. Consequently, simple, direct measurement of soil water and solute flow remains a challenge.
Plant Water Status
Humans have been attempting in some fashion to estimate the water status of plants ever since they discovered that yields could be improved with irrigation. The earliest reported scientific observations of plant water status were probably those of Stephen Hales (1727), whose ingenious experiments on plant water relations are described in detail by Kramer (1949). Hales made sap gauges by attaching glass tubing to the cut ends of branches and roots, from which he obtained observations sufficient to deduce a wealth of information about the flow of water in plants. Progress has been fitful since. Over the past 30 yr or so, a number of elegant tools have been developed for measuring the equilibrium thermodynamic state of water in plant tissue (Boyer, 1995), but these have not been widely used in any field diagnostic sense or as the basis for making management decisions. The natural environment in which plants operate is simply not conducive to equilibrium measurements. Insolation, temperature, atmospheric humidity, and wind speed all can affect plant water status or water potential measurements, and all are inherently unsteady.
Even if plant water potential measurements could be made accurately and routinely in the field, the value of the data is uncertain. As Sinclair and Ludlow (1985) pointed out in the eloquently titled "Who Taught Plants Thermodynamics?", many of the physiological processes that go on within plants, though water dependent, are not well correlated with water potential. Passioura (1988) elaborated on this, noting the false comfort with which plant physiologists had embraced the apparently unifying concept of water potential. As he noted, one of the principal appeals of water potential, namely that it is the driving force for water flow, does not even apply in plants, where solute effects and the properties of membranes complicate matters. What then to measure? There probably is no single answer. Physiologists studying the mechanisms of drought tolerance need different information than growers who are attempting to optimize production of an irrigated crop. Hence, if we are compiling a wish list, our request might be a broader suite of measurement tools.
One potentially valuable and seemingly viable approach might be the use of dielectric methods to measure relative water content. The permittivity of water is so large compared with other plant constituents that the apparent dielectric constant of a leaf, KL, should be a strong function of water content. The challenge is to develop a sensor configuration capable of accurately measuring KL without unduly affecting leaf physiology. Some preliminary research has been done by Ferre and Livingston (personal communication, 2001) using remote shorting-diode waveguides on printed circuits, and it has also been suggested that the measurement could be made in the frequency domain with a flexible etched circuit waveguide (G. Campbell, personal communication, 2001), but a working system has not yet been described in print.
Plant and Soil Nutrient Status
The mantra of precision agriculture is "providing what is needed (and presumably no more), when it's needed, and where it's needed." It sounds simple, but it implies a measurement capability that simply doesn't exist currently. Potassium and P are sufficiently stable and immobile under most conditions so that annual fertilization guided by soil testing is generally adequate, but N is another story. Nitrogen exists in a variety of forms, only two of which can be used by plants, and it is continually being transformed both organically and inorganically. Strategies to reduce nitrate contamination of groundwater by metering its application to match demand require a means to assess either the current N content of a crop or the N-providing capability of the soil to which the plant has access, and to do so rapidly and reliably at multiple points in a field. Most attempts to measure crop N status have been colorimetric, optically detecting the greenness of a leaf or a canopy. The measurement can be made with hand-held devices, or it can be remotely sensed. It is really a chlorophyll detection method, and indeed the hand-held instruments are usually referred to as chlorophyll meters. As a management tool, best results have been obtained when the measurement is made differentially, i.e., referencing crop greenness measurements to measurements made at the same time on a well-fertilized reference strip (e.g., Eghball and Power, 1999). The correlation between chlorophyll content and leaf N content is generally high, but by the time plant N levels have dropped enough to affect greenness, it is often past the point at which N should have been applied. Another way of looking at this is that plant-based measures have no predictive power; they cannot indicate how much N is available at each point in the field for future crop needs.
A method for measuring and forecasting total plant-available soil N across the landscape is probably not a realistic goal, in large measure due to the aforementioned complexity of the N cycle. A more modest, and possibly attainable, goal is a method for measuring and mapping soil nitrate concentrations. A preliminary system was reported by Adsett et al. (1999) that included a soil sampler, conveyor, extraction unit, and nitrate electrode, all mounted on a tractor and controlled by a microcomputer to continuously process samples at regular intervals. Individual components worked well in laboratory tests, but they encountered a variety of operational problems during field tests of the entire system. Presumably others are working on this problem, perhaps with completely different approaches, but at present there is no viable means for a rapid, agronomically useful field measurement of soil N status.
|
CONCLUDING REMARKS |
|---|
|
TOPABSTRACTINTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSURFACE-ATMOSPHERE EXCHANGECONCLUDING REMARKSNOTESREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. L. Steiner and J. L. Hatfield Winds of Change: A Century of Agroclimate Research Agron. J., May 7, 2008; 100(Supplement_3): S-132 - S-152. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
