| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a University of Wisconsin 1525 Observatory Drive Madison, WI 53706
b University of Arkansas 115 Plant Sciences Building Fayetteville, AR 72701
c USDA-ARS University of Minnesota 1991 Upper Buford Circle St. Paul, MN 55108
(jmnorman{at}facstaff.wisc.edu)
Received for publication November 16, 2004.
Dear Editor:
Dr. Gee has made some useful comments regarding measuring water and solute fluxes with tension lysimeters. We agree with some of his comments and disagree with others. We will attempt to briefly address his comments here.
In his first sentence he refers to "the so-called AETL." We believe that this device can be legitimately called an automated equilibrium tension lysimeter even though it has some limitations related to long-term tension measurements in soil; namely limitations of heat dissipation sensors. We do not believe that the major contribution of the paper is the level measurement with the ECH2O probe, although that certainly is significant. In our minds, the major contribution is a robust (has operated for years without trouble) automated tension control system, in contrast to the original system, in which tension was manually adjusted.
We do believe that the AETL design is superior to the PCAPS design. Adjusting the tension in the lysimeters every 10 min to maintain equilibrium with the bulk soil simply is superior to a fixed tension. Dr. Gee wrongly states that the "the heat dissipation units (HDUs) used by the AETL system for sensing tensions have a bubbling pressure typically above 100 cm." The HDUs we use from Campbell Scientific (Logan, UT) have bubbling pressures from 30 to 50 cm of water (3 to 5 kPa); typically we select the units and calibrate each one individually and choose those sensors with the lowest bubbling pressures and most consistent calibrations. After calibrating more than 50 HDUs, we have never observed a bubbling pressure as high as 100 cm of water (10 kPa). Clearly, a bubbling pressure of 30 to 50 cm of water (3 to 5 kPa) is much better at matching conditions in the soil at wetter and dryer tensions than a capillary wick approximately 1 m in length (approximately 100 cm [10 kPa] fixed tension).
The issue of hysteresis in the HDUs is likely not as serious an issue as the mismatch of PCAP tension and soil tension, because even a hysteretic sensor will be closer to the soil tension than a PCAP, except near the fixed tension of the PCAP. Although hysteresis does exist in HDUs, its presence is not as problematic as it could be. We have observed that at the depth in fine-textured soil we routinely work with, namely 1.4 m, soil usually wets up much faster than it dries out; rarely does the soil not wet up to near saturation because major rain events are required for water to get to a 1.4-m depth, even with macropores. During the off-season, the soil remains within 50 cm of water (5 kPa) of saturation for months at a time. Thus, the drying curve is most important if matching is to occur between soil tension and lysimeter tension. We always calibrate HDUs on the drying curve beginning from saturation, just as the soil usually does. This minimizes the hysteresis error. We note here that engineers at Campbell Scientific suggested wetting HDUs under vacuum to be sure all the pores were filled. We learned that this should not be done. The reason is that when the HDU cycles between wet and dry conditions, after the first cycle following vacuum infiltration, the HDU follows the nonvacuum infiltrated response. This anomaly associated with vacuum infiltration could be mistaken for a large hysteresis, which is absent if the HDUs are wet up by simply placing them in water overnight.
The comment by Dr. Gee that Drain Gauges "have collection systems that are superior in design to the proposed drainage-collection system for AETL units" will be true if ECH2O probes are used in the lysimeters, because such probes have errors near 0.3 mm. However, we usually collect the volume of water regularly from the lysimeters and measure it to within a few milliliters so the accuracy is near 0.01 mm over weekly intervals. The automatic recording of the Drain Gauge is a convenient advantage for drainage, but we are also interested in solute concentrations so sample aliquots are required anyway. For rapid sampling we now use a device that sips water out of the lysimeter under suction every 10 min to monitor water flow (in the 0.01-mm accuracy range) and sequentially sample solute concentrations automatically. This system is similar to Kosugi and Katsuyama (2004), who identify the advantages of ETLs: "Such a controlled-tension lysimeter seems to be the most accurate alternative to methods traditionally used to measure water and convective chemical fluxes in the vadose zone." Lentz and Kincaid (2003) also support this view. To us the collection accuracy is not a major issue anyway, because the major problem with tension lysimeters is obtaining representative samples of the drainage and leaching fluxes; that is, how representative is a measurement at one location of the drainage flux over a larger area, even if that measured flux is perfectly accurate.
Dr. Gee's assertion that Fig. 3 demonstrates oversampling of drainage because drainage is larger than precipitation is unlikely to be true. This only occurred at the beginning of the data collection when precipitation before that time is not included. Clearly delays occur at depths of 1.4 m so that precipitation and drainage at 1.4 m are likely to be out of phase. Thus larger precipitation events before the beginning of this data recording period are a better explanation for this initial apparent anomaly. The adequacy of the water balance with ETLs was demonstrated in the paper by Brye et al. (2000) using data over several years.
We believe that the two most serious problems with tension lysimeters in fine-textured soils are obtaining representative samples and disturbing the flow regime above the lysimeters. Because of this we built lysimeters that spanned an entire row spacing of 76 cm below an undisturbed layer of soil. The presence of furrows, stem flow, and preferential flow characteristics can cause heterogeneous flow patterns that may be related to crop management with row crops. Thus, lysimeters smaller than a row spacing are likely to have greater sampling errors than lysimeters that span a row, even at a 1.4-m depth. Clearly Dr. Gee does not agree with us that sampling is one of the most important issues with tension lysimetry; otherwise he would have suggested to Decagon Devices that they make a larger unit since the "approximately 20-cm diameter has been simply a convenience rather than a particular restriction." We worked long and hard to develop a system for reliably installing the relatively large AETLs, having designed and built several devices to shave the contact surface before finding a successful design; therefore, we do not view their installation in undisturbed soil as "an art form." We have successfully installed 14 of these units in Wisconsin, 6 in Minnesota, and 6 in Arkansas using the same procedure without failure; all but one (an HDU failure) of these units are still operating and six have operated for more than eight years continuously. Masarik et al. (2004) stated that the PCAP design by Gee et al. (2002) was based on a disturbed soil above the lysimeters. This comment arises from a description on the Decagon Devices website that the unit could be installed in an auger hole, where the soil on the top of the device clearly would have to be disturbed. Furthermore, in Gee et al. (2002) the same statement about installation in an auger hole is prominent. Because of the length of the Drain Gauge, it appeared to us that installing this device below an undisturbed core would be very difficult. However, we do agree with Dr. Gee that it is possible, "in principle."
Dr. Gee mentioned in his letter that the Drain Gauge has "been installed in undisturbed soils in a side-by-side comparison with AETLs." One of these installations was done in Minnesota and a water balance estimate was compared to two AETL units and a Drain Gauge. The drainage results were as follows: water balance, 135 mm; AETL 1, 118 mm; AETL 2, 76 mm; Drain Gauge, 274 mm. The results indicate that the AETLs are 44 and 13% under the independent water balance estimate and the Drain Gauge is 103% over the water balance estimate.
We do agree with Dr. Gee on several issues; namely, the AETL is a more complex and expensive system ($2000 per lysimeter complete) than the Drain Gauge, and that the Drain Gauge and AETL are likely to be insignificantly different when installed in sandy soils. With regard to sandy soils, both AETLs and the Drain Gauge are easier to install because disturbing the upper layer of soil may cause minimal errors with granular structure; however, heterogeneous patterns of drainage have been observed in sandy soils below row crops of potatoes (Dolan, 1995) because of the hilling nature of the surface. In the Dolan (1995) study drainage and N leaching were twice as high in the furrow as in the row for potatoes grown in Plainfield sand. In this case, lysimeters that span a row spacing clearly will be advantageous. The Masarik et al. (2004) paper dealt entirely with fine-textured soils, and we continue to believe that AETLs have superior accuracy compared to PCAPS for most applications; whether the additional cost and effort is worth the improved accuracy remains to be determined by the user.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal |
