WETLANDS, Vol. 21, No. 4, December 2001, pp. 614–628 䉷 2001, The Society of Wetland Scientists

ESTIMATING EVAPOTRANSPIRATION IN NATURAL AND CONSTRUCTED WETLANDS R. Brandon Lott1 and Randall J. Hunt2 U.S. Geological Survey—Water Resources Division 8505 Research Way Middleton, Wisconsin, USA 53562 E-mail: [email protected] 1 Present Address: Farnsworth Group, Inc. 2709 McGraw Drive Bloomington, Illinois, USA 61704 E-mail: [email protected]

Author to whom correspondence should be addressed

2

Abstract: Difficulties in accurately calculating evapotranspiration (ET) in wetlands can lead to inaccurate water balances—information important for many compensatory mitigation projects. Simple meteorological methods or off-site ET data often are used to estimate ET, but these approaches do not include potentially important site-specific factors such as plant community, root-zone water levels, and soil properties. The objective of this study was to compare a commonly used meteorological estimate of potential evapotranspiration (PET) with direct measurements of ET (lysimeters and water-table fluctuations) and small-scale rootzone geochemistry in a natural and constructed wetland system. Unlike what has been commonly noted, the results of the study demonstrated that the commonly used Penman combination method of estimating PET underestimated the ET that was measured directly in the natural wetland over most of the growing season. This result is likely due to surface heterogeneity and related roughness effects not included in the simple PET estimate. The meteorological method more closely approximated season-long measured ET rates in the constructed wetland but may overestimate the ET rate late in the growing season. ET rates also were temporally variable in wetlands over a range of time scales because they can be influenced by the relation of the water table to the root zone and the timing of plant senescence. Small-scale geochemical sampling of the shallow root zone was able to provide an independent evaluation of ET rates, supporting the identification of higher ET rates in the natural wetlands and differences in temporal ET rates due to the timing of senescence. These discrepancies illustrate potential problems with extrapolating off-site estimates of ET or single measurements of ET from a site over space or time. Key Words:

evapotranspiration, water budget, PET, lysimeters, hydrology, geochemistry

INTRODUCTION

Kadlec 1986, Mitsch and Gosselink 1993). Proper identification of the sources and sinks of water, and the processes by which water and associated constituents move through the system, is essential to calculating nutrient, energy, and chemical budgets (LaBaugh 1986). An understanding of the wetland water budget and its associated chemical budget is also needed to characterize functions associated with wetland systems accurately. The success of compensatory mitigation efforts is dependent on the ability to characterize the on-site hydrology (National Cooperative Highway Research Program 1996). Improved characterization of the hydrology of a site may help to increase the success rate of

Increasingly, wetlands are being recognized as integral parts of the landscape, performing functions necessary to maintain a healthy environment. Recognition of their value to society has heightened awareness of the effects of wetland loss and fueled the adoption of laws attempting to protect wetlands (National Cooperative Highway Research Program 1996). As a result, on-site, in-kind mitigation has been recognized as essential to the ‘‘no net loss’’ objective in the United States established by the National Wetland Policy Forum (Conservation Foundation 1988). In order to meet these goals, understanding the hydrology of wetland systems is critical (Carter 1986, Hammer and 614

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION such wetland mitigation projects, which often do not meet their design criteria (Kusler and Kentula 1989). Accurately characterizing wetland hydrology, however, can be difficult (National Research Council 1995). Water-budget calculations are typically associated with large errors due to errors in the calculation of the terms in the water balance (Dooge 1972, McGuinness and Bordne 1972). Improvements in the ability to characterize one component of the water budget will improve the overall budget, as well as analyses derived from the water budget. Energy budgets have been long recognized as the most accurate way to estimate ET (Harbeck et al. 1958) but are rarely performed in practice due to the considerable investment of instruments and personnel required (Sturrock et al. 1992). Because of difficulties in its direct measurement, evapotranspiration (ET) is often estimated as a residual term (Winter 1981, LaBaugh 1986) or calculated using off-site data (e.g., Yin and Brook 1992, Bravo and Brown 1998). When on-site data are collected, analyses are commonly restricted to relatively simple meteorological equations (e.g., Hammer and Kadlec 1986, Yin and Brook 1992, Mitsch and Gosselink 1993). Recent studies have begun to evaluate estimates of ET in natural wetland systems (e.g., Koerselman and Beltman 1988, Lafleur and Rouse 1988, Lafleur and Roulet 1992, Price 1994, Abtew 1996, Campbell and Williamson 1997) and disturbed wetlands (e.g., Souch et al. 1998, Thompson et al. 1999), although sparse attention has been paid to constructed wetlands. As a result of these investigations, the evapotranspiration term in vegetated wetlands has become the subject of debate; studies have shown that the presence of wetland vegetation can result in either higher or lower rates of evaporation compared to open water (e.g., Idso 1981, Idso and Anderson 1988, Abtew and Obeysekera 1995). This paper presents the results of a field study of natural and constructed ground-water-dominated wetlands. The primary objective of this study was to compare potential evapotranspiration (PET) estimated by commonly used meteorological approaches with estimates of evapotranspiration determined by other direct (and more labor-intensive) methods such as monitoring non-weighing lysimeters and water-table fluctuations. The focus was not to develop a modified meteorological equation for PET, as this approach often has limited transferability beyond the site studied. Rather, it is to evaluate the utility of a simple meteorological approach that is often used in compensatory mitigation projects to characterize spatial and temporal differences in ET. A secondary objective was to use insight gained from small-scale, root-zone chemistry to assess the validity of the ET measurements. This is, to our knowledge, one of the first works to use geo-

615

Figure 1. Site map showing the locations of the instrumented site in the natural wetland (W1) and the site in the constructed wetland (F2). Contour elevations are in meters above mean seal level.

chemical methods to assess wetland ET rates in a ground-water-dominated wetland and one of the first to compare different techniques for measuring ET at a natural and constructed wetland directly. The ability to understand not only the rate, but also the variability of water lost to evapotranspiration in wetlands, should enable scientists and managers to measure wetland hydrology more accurately, which in turn may facilitate mitigation success. SITE DESCRIPTION The Wilton wetland complex is located along the Kickapoo River in Monroe County, in the unglaciated region of southwestern Wisconsin, USA (Figure 1). This area typically receives 81 cm of precipitation per year (Linsley et al. 1982) and received 79 cm during the year studied (1996). Hughes et al. (1981) describe the Kickapoo River basin as having steep slopes (30 to 40 percent), rounded ridges, and flatter alluviumfilled valleys. Geology of the site consists of thin (⬍5 m) fluvial/lacustrine sediments in the river bottoms; these sediments were derived from the Cambrian and Ordovician bedrock and overlie sandstone bedrock of Cambrian age. Adjacent bluffs are comprised of Cambrian sandstone capped by Ordovician carbonates. Natural wetlands within the floodplain have thin accumulations of peat (ranging from 0.1 to 1 m) overlying fluvial deposits. The peat contains varying

616 amounts of silt, influenced by agricultural practices in the adjacent highlands. Hunt et al. (1999b) reported that the natural wetland consists of one hectare of shrub-sedge meadow dominated by Carex tricocarpa Muhl. and Carex stricta Lam., with a prevalence of Eupatorium maculatum L., Aster puniceus L., Impatiens capensis Meerb., Galium asprellum Michx., Leersia oryzoides (L.) Swartz., Lycopus uniflorus Michx., Polygonum sagittatum L., and Alnus incana ssp. rugosa (L.) Moench. The herbaceous layer reached a height of approximately 1.5 m during the growing season, and the woody layer was approximately 2–3 m in height during the study period. The root zone of the natural wetland was approximately 30 cm deep. In addition to the shrub-sedge meadow wetland, there is also an adjacent 7.5 hectares of riparian wetland dominated by Alnus incana, Ulmus americana L., and Fraxinus nigra Marshall that was not investigated during this study. During the summer of 1991, 3.8 hectares of adjoining upland agricultural field (shown as the constructed wetland in Figure 1) were excavated as part of a project to mitigate sedge meadow loss associated with an adjacent road project. The field was excavated to a maximum depth of two meters based on pre-construction water levels from 72 wells in the field and in the natural wetland. Salvaged marsh surface (SMS) from the destroyed wetland (0.4 hectare) was applied over the constructed wetland. Additional SMS was obtained from an off-site wetland to complete the project. The constructed wetland was characterized by a plant community consisting of Juncus effusus L., Setaria glauca (L.) P. Beauv., Phalaris arundinacea L., Eupatorium perfoliatum, Aster ericoides L., Verbena hastata L., Carex vulpinoidea Michx., and Juncus tenuis Willd. (Hunt et al. 1999b). The constructed wetland consisted of only an herbaceous layer that reached a height of approximately 1 m during the growing season and had a root zone approximately 15 cm deep. Surface-water inflow and outflow are minimal components of the water budget for both the natural and constructed wetlands as a result of the geomorphic setting (wetlands located in flat valley bottom), site hydrology (ditches that separate the wetlands from the upland slopes), and construction design (berm constructed between the river and the constructed wetland). The physical hydrogeology of the Wilton wetland was previously investigated with an intensive network of nested wells that were monitored for eight years, continuous water-level monitoring (Hunt et al. 1999a), areal two-dimensional analytic element-flow modeling (Hunt 1992), three-dimensional ground-water-flow modeling (Bravo and Brown 1998), and stable isotope mass balance, temperature-profile modeling, and numerical water-balance modeling (Hunt et al. 1996).

WETLANDS, Volume 21, No. 4, 2001 This work has shown that the site lies within a regional ground-water-discharge area and is characterized by strong upward gradients (average 0.1 m/m and locally ⬎0.2 m/m). The magnitude of ground-water discharge varies spatially but generally is characterized by increasing ground-water discharge near the Kickapoo River (Hunt et al. 1996). Hunt et al. (1996, 1997, 1998, 1999a, b) describe two stations in the natural and constructed wetlands that were established for instrumentation (sites W1 and F2 in Figure 1). These sites are approximately 300 m apart and considered representative of the surrounding wetlands. At both sites, ground-water discharge is a dominant source of water, supplemented by precipitation (Hunt et al. 1998). Hunt et al. (1999a) describe the hydroperiods at the two sites. Water levels are below land surface during the majority of the growing season; the median depth to ground water over the growing season is similar at the two sites (9.9 cm for W1 and 7.9 cm for F2), as were the standard deviation of the measured hourly water levels (1 standard deviation ⫽ 12 cm at W1 and 14 cm at F2). Evapotranspiration has been shown to be the dominant sink during the growing season (Lott 1997). Although the natural and constructed sites are close in proximity and are supplied by the ground-water system, there are notable differences. The majority of shallow substrate at the natural wetland has developed in situ as the result of deposition of sedge detritus; silt and fine sand derived from adjacent upland agriculture also are present in the soil column. The dry bulk density of this natural, fibric peat is 1.5 g/cm3. At the constructed wetland, a thin layer (15 cm) of the offsite salvaged marsh surface placed over mineral soils during construction of the mitigated wetland yielded a soil with a dry bulk density of 1.75 g/cm3. In addition, hourly wind speeds measured during 1995 and 1996 (n ⫽ 282 days) were appreciably greater and more variable at the constructed wetland F2 site (average ⫽ 3.1 km/hr, standard deviation ⫽ 1.8 km/hr) than at the natural wetland W1 site (average ⫽ 1.9 km/hr, standard deviation ⫽ 1.1 km/hr). METHODS Meteorological Estimation of Potential Evapotranspiration (PET) Because a supply of water is seldom considered limiting during the growing season in many types of wetlands (Mitsch and Gosselink 1993), the theoretical atmospheric demand or potential evapotranspiration (PET) is used to approximate wetland evapotranspiration rates. Lott (1997) describes evaluation of three meteorological methods (Thornthwaite, Penman com-

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION bination, and Priestley-Taylor) of estimating potential evapotranspiration at sites W1 and F2 (Figure 1). No appreciable differences in the PET rates calculated by these methods were observed at this site during the two years studied. Therefore, only the Penman combination method is included here for brevity. The Penman combination method requires temperature, relative humidity, wind speed, and net radiation data measured at 2 m above land surface, as well as a measurement of radiation energy lost to the soil. The method estimates PET as a weighted sum of a rate estimated by an energy balance and a rate estimated by mass transfer. The Penman combination equation is reproduced from Dunne and Leopold (1978) as follows: PET ⫽ ((⌬/␥) (H-G) ⫹ Ea)/((⌬/␥) ⫹ 1)

(1)

where PET is the potential evapotranspiration in cm/ day, ␥ is the psychometric constant in mb/⬚C, H and G are net radiation and soil heat flux, respectively, in cm/day of evaporation calculated by dividing the flux (W/m2) by the latent heat of vaporization, and Ea is the mass transfer evaporation rate in cm/day. The slope of the saturation vapor pressure versus temperature curve (⌬) is given in mb/⬚C by ⌬ ⫽ (4098esa) / (237.3 ⫹ T)2

(2)

where esa is the saturation vapor pressure in mb and T is the air temperature in ⬚C. The mass transfer evaporation rate recommended by Penman (Dunne and Leopold, 1978) in cm/day is given by Ea ⫽ (0.013 ⫹ 0.00016u)*((1-RH)esa)

(3)

where u is the average daily windspeed in km/day, RH is the relative humidity as a decimal fraction, and esa is the saturation vapor pressure for a given T in millibar measured at 2 m above the land surface. The coefficients (0.013 and 0.00016) define an empirical relationship for the mass transfer evaporative losses over well-watered alfalfa, 30–50 cm tall. Weather stations were installed at site W1 in the natural wetland and site F2 in the constructed wetland to obtain the required PET input. Each recording station consisted of a Vaisala HMP35C temperature and humidity probe, an RM Young 101 anemometer, and an REBS Q6 net radiometer located at a height of 2 m above land surface. An REBS HFT3 soil heat flux plate was buried at a depth of 5 cm and a Campbell Scientific 107B soil temperature probe was buried at a depth of 2 cm to estimate losses between the soil heat flux plate and the soil surface. All sensors were properly shielded, maintained, and calibrated, and sensor height was not varied during the study period. The only values used in either calculation that were not

617 collected on site were estimates of the heat capacities of the soil at each station. A value of 48.6 W/m2 was used for the natural wetland peat, and 41.7 W/m2 was used for the constructed wetland salvaged marsh surface. These values were obtained from tables listing soil types and their corresponding heat capacities in Campbell and Norman (2000). Meteorological data were collected from May 19 to October 31, 1995 and from April 15 to October 31, 1996 at the natural wetland. Data were obtained from July 18 to October 31, 1995 and from April 15 to October 31, 1996 at the constructed wetland. The data used in the calculation of PET were recorded by a Campbell Scientific CR10 datalogger every minute, and an average hourly value for each variable was calculated. From those data, a value of PET in cm/day was calculated for each hour. The hourly PET was then totaled for each day. Monthly PET was given as an average of the daily PET calculated. Direct Measurement of Evapotranspiration One of the most important parameters for obtaining direct measurements of ET is specific yield—the ability of the soil to store water. Unfortunately, bulk specific yield can be difficult to quantify because of problems with heterogeneity in the soil column and rapid water-table responses in peat due to inputs of water (O’Brien 1982, Heliotis and DeWitt 1987). A total of 38 measurements of air-filled porosity (for simplicity referred to here as specific yield) were collected at the natural and constructed wetland using methods outlined by Gerla (1992). While there was variation between measurements at each site, the natural wetland had an appreciably greater value than the constructed wetland in almost all measurements (Lott 1997). The average specific yield of the natural wetland (0.15, n ⫽ 19, standard deviation ⫽ 0.06) was about twice the average value measured in the constructed wetland (0.07, n ⫽ 19, standard deviation ⫽ 0.04). To measure evapotranspiration rates directly, a nonweighing lysimeter was installed at both the natural and constructed wetland sites. The lysimeter technique has been considered ‘‘very suitable’’ for the measurement of evapotranspiration in wetlands (Koerselman and Beltman 1988). Each lysimeter consisted of a steel cylinder 35 cm in diameter (area ⫽ 962 cm2) and 45 cm deep. The lysimeters had a sealed bottom and were installed with a 5-cm lip above land surface, limiting the local water budget to precipitation and evapotranspiration by eliminating exchange with ground water and surface water. Care was taken to minimize compaction of the soil during coring and emplacement. Effort was also made to minimize errors generated by the oasis effect (Idso 1981) by making lysimeter mea-

618

WETLANDS, Volume 21, No. 4, 2001

surements in situ and keeping water levels within the lysimeter representative of those measured outside the lysimeter in the surrounding wetland soils. The lysimeters were installed in representative zones of vegetation and were filled with peat and vegetation from the lysimeter excavation. No significant differences between vegetation structure and health were observed between the lysimeter vegetation and surrounding vegetation once the vegetation inside the lysimeter had established. Measurements of water levels inside the lysimeter were collected weekly during the 1995 and 1996 growing seasons using a steel tape and chalk (reported accuracy ⫾ 0.3 cm). At the beginning of each period used for ET calculation, the water levels were measured both inside and outside of the lysimeter. Water was then slowly added to the lysimeter until the water levels measured inside and outside of the lysimeter were equal, and the water levels were recorded. Periods in which no water level could be recorded in the lysimeter piezometer at the end of the analysis period (i.e., the lysimeter was dry) were not included in the analysis, as the water available for ET within the lysimeter was artificially limited beyond that outside of the lysimeter. Evapotranspiration can also be measured directly from the diurnal fluctuation of the shallow water table in wetlands (e.g., Todd 1964, Mitsch and Gosselink 1993, Owen 1995, Rushton 1996). Hourly water levels were recorded outside of the lysimeter at both the natural and constructed site using a capacitance probe and datalogger with a reported accuracy of ⫾ 1 cm. To compare evapotranspiration losses measured from water-table hydrographs to the evapotranspiration rates measured in lysimeters, the continuous hydrograph was divided into multiple week periods matching the lysimeter measurements. A simple water balance was then established for the period using the following equation: ET ⫽ Sy (GW ⫺ ⌬S) ⫹ P

(4)

where ET, P, and GW are the average daily fluxes of evapotranspiration, precipitation, and ground-water inflow, respectively, during the period, Sy is the specific yield for the soil, and ⌬S is the rise (⫹) or fall (-) in water level during the period. The use of longer periods for calculation has the advantage of buffering short-term variability in water levels due to atmospheric variations (i.e., barometric pressure), air entrapment, and capillary zone drainage. Root Zone Geochemical Profiles Stable isotopes have been shown to be useful in estimating hydrologic fluxes in lakes (Krabbenhoft et

al. 1990) and wetlands (Hunt et al. 1996), and stable isotopes of water have been shown to be useful in distinguishing between precipitation and ground-water sources at this site (Hunt et al. 1998). In this study, the stable isotope ratio (18O/16O) from root-zone water samples was used to investigate the differences in residence time of rain in the natural and constructed wetland root zone. In addition to water isotopes, select solutes can be used as conservative tracers in circumneutral (pH near 7) aquatic systems (Stauffer 1985). In this study, the effects of ion concentration due to ET on calcium profiles in the root zone were used to make qualitative assessments of ET rates; plant uptake of Ca was assumed to be negligible during the period studied. By monitoring the solutes and isotopes in the root zone of the wetland, interactions between infiltration of rainfall and ET losses can be compared between sites throughout the growing season. Wetland root-zone isotope and ion profiles were collected with close-membrane equilibrators or ‘‘peepers’’ (Hesslein 1976) at both the natural and constructed sites at three-week intervals throughout both growing seasons. The peepers were installed outside of the lysimeters and sampled using the methodology of Hunt et al. (1997). The isotopic composition of the ambient, deep ground water at the natural and constructed wetland sites had been characterized previously (Hunt et al. 1996, 1998). The composition of bulk rainfall was measured during the 1995 and 1996 growing seasons to ensure that precipitation values were representative of the study period. Isotopic analyses were performed at the U.S. Geological Survey (USGS) National Research Program Laboratory in Menlo Park, California. Oxygen-18 values were measured by CO2 equilibration (Epstein and Mayeda 1953) and introduction into a Finnigan-Mat 251 mass spectrometer. Calcium was analyzed using inductively coupled plasma (ICP) emission spectrometry at the University of Wisconsin Soils and Plant Analysis Laboratory. COMPARISON OF EVAPOTRANSPIRATION ESTIMATES Spatial Differences When daily PET rates are averaged over a given month of the growing season, the results (Figure 2: range 0.14–0.37 cm/d, average ⫽ 0.26 cm/d, n ⫽ 8) agreed with average daily values of ET previously reported in the literature (e.g., 0.15 to 0.31 cm/d reported by Campbell and Williamson 1997), and appreciable differences between the natural and constructed sites were not noted (Figure 2). This is expected, as the stations are 300 m apart and do not differ significantly

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION

Figure 2. Average monthly values of PET for the constructed wetland and natural wetland calculated using all data collected during the growing season. The monthly averages consist of the average of hourly data collected during the period.

in aspect, elevation, or exposure to sunlight; thus, average monthly values of temperature and relative humidity were not appreciably different (Lott 1997). Because the Penman combination PET equation used here is primarily radiation-driven, similarities in the radiation budget between the sites result in similar monthly PET values. Lott (1997) provides a more comprehensive discussion and comparison of meteorological methods used to estimate PET. In the natural wetland, direct measurement techniques (lysimeter and water-table fluctuation) shown in Table 1 and Figure 3 also agreed with other previously published values of ET (e.g., mean daily rates of 0.25–0.69 cm/d reported by Lafleur 1990) and gen-

619 erally compared well with each other (Figure 3), but did not always agree with estimates of PET. Unlike results typically reported in the literature (e.g., Koerselman and Beltman 1988, Lafleur 1990, Lafleur and Roulet 1992, Campbell and Williamson 1997, Thompson et al. 1999), the directly measured ET rates in the natural wetland exceeded PET (as estimated by the Penman combination method). In the constructed wetland, Penman PET rates were comparable to measured rates in the early and middle growing season (Table 1), although periods of notable differences exist (e.g., 5/31/96 through 6/3/96 in Figure 3). Direct measurements of ET from the natural wetland were generally greater than those measured in the constructed wetland (Table 1, Figure 3). Similarly, Lafleur and Rouse (1988) noted that wetlands with woody vegetation had higher ET rates than a nearby sedge community that had a water table below land surface. Penman PET more closely approximates the directly measured ET in the late growing season at the natural wetland site but may slightly overestimate measured ET at the constructed site (Table 1, Figure 3). There are several possible reasons why PET and directly measured ET differ in the natural wetland. Clearly, the actual ET rate cannot exceed a potential ET rate that describes the capacity of the atmosphere to absorb water vapor. Therefore, one possible reason is that the meteorological methods are underestimating the ability of the atmosphere to convey water vapor from the natural wetland. The Penman combination method used here (as are most simple meteorologically derived equations of ET) was developed for agricultural applications and contains simplifying assumptions that may not be appropriate for the wetland systems investigated in this study. For example, the methods often assume a one-dimensional diffusion trans-

Table 1. Comparison of daily average evapotranspiration rates derived from Penman PET, Lysimeter, and Water-Table Well methods in cm/d. Numbers of days (n) and ET rates are from time periods within the growing season when estimates from all three methods were available at both sites. Natural Wetland Site W1

Constructed Wetland Site F2

Penman PET

Lysimeter ET

Water-Table Well ET

Penman PET

Lysimeter ET

Water-Table Well ET

0.26

0.45

0.38

Early (5/1/96–6/30/96) n ⫽ 18 days

0.26

0.34

0.29

0.40

0.56

0.56

Middle (7/1/96–8/30/96) n ⫽ 23 days

0.40

0.35

0.48

0.19

0.26

0.22

Late (9/1/96–10/15/96) n ⫽ 16 days

0.22

0.11

0.12

620

WETLANDS, Volume 21, No. 4, 2001

Figure 3. Comparison of methods for periods of the 1996 growing season at both the natural and constructed wetland sites. Early, middle, and late portions of the growing season are indicated. Periods where data were not available for all three methods are indicated.

port off a vegetation community with a constant height. Other assumptions include advective removal of the high humidity boundary layer through wind transport, again assuming a uniform, average wind speed over a vegetation community of uniform height. This may be inappropriate, as the equation does not incorporate variables directly measured within the vegetation cover itself nor changes in vegetation over the growing season. The heterogeneous airflow likely increases contact with the atmosphere and enhances turbulent flow near the earth’s surface. This turbulence,

in turn, would facilitate exchange between the highhumidity boundary layer and the drier upper atmosphere. The differences in canopy may also explain why the PET estimate is closer to the ET measured in the constructed wetland than in the natural wetland. All Penman variables were collected at 2 m above the land surface; yet this results in a height that ranges from 0.5 to 1.5 m above the canopies of the respective vegetation communities as the growing season progresses. The natural wetland was characterized by slightly tall-

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION er and more heterogeneously sized vegetation (as expected from a wetlands with a herbaceous and shrub layer) than that observed in the constructed wetland. While this difference did not seem to result in appreciable differences in the PET estimates for the two sites (Figure 2), it does point out the different degree to which the two sites fit the underlying assumptions of the meteorological PET equations. The idea of a uniform height for the vegetation community may adequately represent agriculture sites but does not represent the variable height of the natural shrub-sedge meadow plants. These considerations are likely not as important in the constructed wetland with its homogeneous, uniform herbaceous plant heights, which are more appropriate for the assumptions used in the agricultural applications where these methods were developed. Alternatively, it could be argued that the results are an artifact of uncertainty associated with the lysimeter and water-table direct measurements of ET. There could be an oasis effect that effects the lysimeter-derived direct ET measurements (thus overestimating actual ET rates in the natural wetland) that was not included in PET measurements. However, direct measurements of the water-table hydrograph outside of the lysimeters and geochemical analyses (discussed below) provide evidence for higher, spatially variable rates outside of the lysimeters not represented in the meteorological PET measurements shown in Figure 2. There is also appreciable uncertainty with the specific yield parameter used in the water-table method. Because it is a difficult parameter to quantify, it could be argued that errors in specific yield might be responsible for the difference between the water-table/lysimeter measurements and the meteorological estimate. However, the values reported for the wetlands (0.15 and 0.07 for the natural and constructed wetland soils, respectively) are lower than what is often reported for wetlands. That is, if there is a bias, it might be expected that the actual specific yield is greater than that used in the direct measurement using the water-table method. A greater specific yield would result in a higher value of ET calculated by the direct measurements; this, in turn, would increase the discrepancy between the Penman PET and direct ET measurements observed at the natural wetland site. There likely are other meteorological equations that might better represent the natural wetland studied here. For example, equations like the Penman-Montieth equation or one by Sheffe (1978) include local measurements of canopy structure and roughness that may provide more accurate representations of site-specific PET by reflecting the differences in advective transport capacity governed by the differences in windspeed. In addition to spatial heterogeneity within the system,

621 Owen (1993) suggested that temporal variations (e.g., short term variations in humidity and windspeed) could be important for accurate meteorological estimates of PET. These factors could be accounted for by using a more sophisticated energy balance approach (e.g., Bowen’s ratio method, eddy correlation method). The additional data and fetch distance needed to perform these more sophisticated techniques, however, complicate simple estimates of water balance. These lines of inquiry are beyond our objective of comparing a commonly used simple method of estimating ET to more labor-intensive direct measurements and remain a topic of further research. Temporal Differences The degree to which estimates of ET using simple meteorological methods agreed with estimates of ET using direct measurement methods varied over time in both the natural and constructed wetland. While all methods followed the same general trend, with ET peaking in late July and decreasing as plants senesce, the ability of simple meteorological methods to represent the field measured ET varied. As noted above, Penman PET was lower than measured ET rates in the natural wetland during the height of the growing season (Figure 3). During this same period, ET rates measured in the constructed wetland are often near calculated PET rates, suggesting that the constructed wetland ET had reached the potential evapotranspiration rate calculated with the Penman combination equation. Late in the growing season, the constructed wetland lysimeter ET was consistently lower than the Penman PET, while the natural wetland lysimeter ET tracked the Penman PET more consistently (Table1). One possible reason for the late-season discrepancy is the timing of plant senescence. Others have noted that wetlands can have different ET rates early in the growing season before leaf out (non-vegetated) and later in growing season after leaf out (e.g., Lafleur and Rouse 1988, Lafleuer 1990). However, there is likely a third segment of the ‘‘growing season’’ that may be important—the period after parts of the plant community senesce. Field observations indicated that, in September, the grass-dominated vegetation in the constructed wetland began to senesce (USGS unpub. data 1996). Subsequently, ground-water levels increased to land surface and precipitation accumulated because the evaporation was not sufficient to remove the inputs of water. The sedge-dominated vegetation in the natural wetland did not show this behavior until mid-October. This finding suggests that wetland plant transpiration, and the wetland plant morphological adaptations that enhance transpiration,

622

WETLANDS, Volume 21, No. 4, 2001

Figure 4.

Water-table hydrograph measured in natural wetland from 8/20/96–9/01/96.

can be an important factor in estimating wetland ET rates. The wetland plant community can have an extraordinarily large effect on the wetland water budget. Others have noted that some plants (e.g., Typha latifolia L., Phragmites australis (Cav.) Trin. Ex Steud., Nymphaea ordorata Ait, Menyanthes trifoliata L.) contain a convective-flow system that can actively pump air and water vapor from the root zone to the atmosphere (e.g., Dacey 1981a, 1981b, Grosse et al. 1991, Armstrong and Armstrong 1991). One mechanism of flow reported for emergent macrophytes states that flow originates in the living leaf sheaths and nodal stomatal regions; convected gases are transported via air spaces through the plant stem and underground rhizome and are ultimately vented through old, broken culms (Armstrong and Armstrong 1991). The rate of flow follows a diurnal cycle (Chanton and Whiting 1996) and is enhanced by environmental conditions such as low relative humidity and Venturi convection resulting from high wind speeds (Armstrong et al. 1996). Flow velocities of up to 800 mm/min and volumetric flow rates of 16 ml/min per shoot have been recorded in the field (Armstrong and Armstrong 1991). The convection rates are reduced during plant senescence due to the reduction in living sheaths available for driving the system. We do not know to what extent the plant communities in the wetlands at our site use convective flow adaptations due to the small number of plants where convective flow mechanisms have been investigated. It is conceivable that appreciable flow is occurring at

the site. If true, the system would be expected to switch from active pumping to a more passive ET mechanism during senescence; water would be obtained from the ground surface or through diffusion up the plant stems, and resulting ET rates will be largely constrained by the high humidity, low wind exchange areas present in the standing vegetation. The type of plant species present in the community, its density, and time of senescence, would affect the overall transpiration of the wetland system. Moreover, all three factors would not be included in simple meteorological measurements of PET commonly used in wetland studies. The effect of plant activity on the short-term ET rates can also be demonstrated using a two-week period in the growing season before senescence. A hydrograph from the natural wetland (Figure 4) shows two slopes in the recession after a 2.2-cm rain, suggesting two different ET rates if we assume that the specific yield is homogeneous in the profile. We can attribute the decrease to transpiration rather than evaporative drying because water isotopes are not fractionated (see Hunt et al. 1998). The slope of the watertable decline is 6.2 cm/day while the water table is in the area of highest root density; as the water level drops below the area of high root zone density, the slope of the water-table decline decreases to 1.4 cm/ day. The differences in slope were largest during hot, dry periods when water table is in the root zone (Lott 1997). These periods of enhanced ET rates are measured in the water table and result from a different process than the more frequently cited enhanced ET

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION rates that result from wet canopies (e.g., Lafleur 1990, Thompson et al. 1999). Wet canopies affect the amount of water evapotranspired to the atmosphere but would not effect the slope of the water-table decline, as the additional water lost to the atmosphere is derived from interception of precipitation rather than ground water transported from the root zone. The efficiency of water transport in the ground to the plant roots is a likely driver for the observed change in slope. During the time when the roots obtain water directly from the saturated zone, there is efficient conveyance of water. During periods when the water table is below the root zone, on the other hand, the roots obtain water through capillary forces. Because the plant roots must overcome tension forces in the capillary fringe to obtain water, they would be expected to be less efficient in achieving potential evapotranspiration. The loss of efficiency as distance to the root zone increased has also been noted by others (e.g., Todd 1964). Differences in efficiency of transport may also help explain the difference between the constructed and natural wetland ET rates—the constructed wetland root zone is shallower (15 cm below land surface) than the natural wetland (30 cm below land surface). While growing season water-level statistics are similar between the two sites, the water table is generally more flashy and farther below the surface at the constructed wetland (e.g., the hydrographs shown in Figure 5). In any event, differences in the ability of the soil to transmit water suggest a re-evaluation of the commonly held assumption that wetland plants are seldom water-limited may be warranted, especially for ground-water-dominated systems where the water table is commonly below land surface. GEOCHEMICAL INVESTIGATION In the discussion above (and noted by Penman, among others), it was shown that Penman PET might not predict the evapotranspiration rates measured in the field consistently. That is, while the Penman PET rates calculated at the natural and constructed sites did not differ significantly, the field ET rates measured at the respective sites showed appreciable differences. In addition, ET rates appear to be related to plant senescence; thus, differences in plant communities between the natural and constructed wetland resulted in different ET rates late in the growing season. While these observations identify potential inadequacies in some simple meteorological equations, additional investigation using independent methods would help support these observations. We used geochemistry to elucidate the discrepancy between the meteorological and directly measured values of ET and the apparent tem-

623 poral differences in ET due to plant senescence qualitatively. Spatial Differences Hunt et al. (1996 and 1999a) have documented the strong vertical flow present at the W1 and F2 during the growing season. If there is sufficient separation in the isotopic signature of ground water and precipitation sources, stable isotopes of the water allow us to track the vertical movement of these sources through the root zone. If the sites, 300 m apart, have the same ET rate and infiltrate the same amount of rainfall, the average value of the isotope profile within the root zone should be similar. However, if more rainfall in the natural wetland root zone has been removed by a higher ET rate, the rainwater should be less prevalent in the profile. This case is illustrated for early July 1996 in Figure 5. Appreciable rainfall occurred periodically during the period prior to the removal of the peeper from the wetland root zone. At this time of the growing season, rainfall has a ␦18O composition of ⫺5 ‰, and the ambient ground-water ␦18O composition is typically ⫺9 to ⫺9.5 ‰ (Hunt et al. 1996, Lott 1997, Hunt et al. 1998). As shown in Figure 5, the deeper sections of the root-zone isotopic profiles in both the natural and constructed sites are dominated by the water with a ␦18O composition more similar to ground water. However, in the uppermost portions of the profile, the constructed wetland’s root zone porewater is primarily rainwater. The isotope profiles obtained from the natural wetland, from a sampler emplaced over the same period of time, had less rainwater than profiles collected in the constructed wetland profiles. One might argue that the profiles result from unequal amounts of rain infiltration at the two sites. We do not believe this to be the case because 1) we have not observed significant overland flow at the site during seven years of study and 2) the natural wetland has greater porosity that would facilitate the infiltration of rain. It might also be argued that differences in specific yield will result in different profiles. While this is undoubtedly the case at these two sites, given the difference in specific yield, the uppermost water at the natural wetland should still consist of 100% rainwater in a distinct, albeit thinner, layer. Such a layer was not observed. Thus, we conclude that less rainwater was present in the natural wetland root zone because the natural wetland vegetation evapotranspires at a higher rate than the constructed wetland plant community. Temporal Differences We also used solute geochemistry to gain insight into the apparent changes in ET rates observed in the

624

WETLANDS, Volume 21, No. 4, 2001

Figure 5. Isotope profiles of root zone porewater. Data from peepers installed in natural and constructed wetlands from 6/ 26/96 to 7/17/96. Rainfall and respective hydrographs are also shown.

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION

625

Figure 6. Root-zone porewater calcium profiles. Two sets of peepers installed in both the natural and constructed wetlands on 8/08/95. One set of peepers was removed from each wetland on 8/30/95; the second set was removed one week later.

natural and constructed wetland. Two peepers were installed in both the natural and constructed wetlands. One peeper was removed from each wetland on August 30, 1995. The second peeper at each site was left in the wetland for an additional week and removed on September 6, 1995. A rainfall event occurred during the three-week period in which both peepers were in the wetland; no rain fell during the week between the collection of the first and second peeper (August 30– September 6, 1995), and the weather was hot and dry (Lott 1997). Similar to the water isotope discussion, it is expected that the solute profiles will be similar in the natural and constructed wetlands if the ET rates are similar. As shown in Figure 6, this is not the case. In the week that the second peeper was emplaced, the natural wetland solutes became more concentrated, as would be expected in a system where plants are actively evapotranspiring. However, the constructed wetland solute profile became more dilute during this same period of time, indicating that minimal ET is occurring and the constructed wetland plants have likely entered senescence. Similar to the isotope data, these data suggest that the two wetland systems have different ET rates; Penman PET rates calculated during this time period did not indicate different PET rates between the sites.

The fact that the constructed wetland becomes more dilute during a period that no rain fell is an artifact of the 3-week emplacement required for the peepers to equilibrate. That is, the peepers are still equilibrating to the stress of the rain that occurred during August 16–28, 1995. While this equilibration lag time precludes the quantification of ET rates, the fact that the peepers in the natural and constructed wetlands were of identical construction and emplaced during the same period allows their use to assess qualitative differences between the natural and constructed wetland root zones.

IMPLICATIONS FOR WETLAND RESTORATION AND CREATION The results of this study reveal differences in hydrologic flux from the natural and constructed wetlands. While the PET measurements suggested that atmospheric demand and the availability of water in the root zone were essentially equal between the two sites, significant differences in the rate and seasonal variability of field evapotranspiration were identified. The observations that PET rates may not represent overall site ET and that actual site ET can vary between wetlands 300 m apart underscore potential problems with

626

WETLANDS, Volume 21, No. 4, 2001

Table 2. Comparison of cumulative growing season estimates of evapotranspiration and potential evapotranspiration at the natural and constructed wetland sites (cumulative evapotranspiration in centimeters for 5/1/96 through 10/15/96 assuming early-middlelate average daily ET rates calculated using times when estimates from all methods were available are representative of the entire early-middle-late period).

Natural Wetland Site W1 Constructed Wetland Site F2

Penman PET

Lysimeter ET

WaterTable Well ET

48.0 49.3

72.0 46.2

66.1 51.5

planting plans and wetland operation plans—especially in treatment wetlands where shifts in the water balance can have dramatic shifts in the water quality of the effluent (Kadlec and Knight 1996). Moreover, an understanding of how different plant species evapotranspire could improve the design of created wetlands. For example, if larger flows were expected later in the growing season, the design should specify plants that senesce later instead of the wetland grasses or forbs found in this constructed wetland. SUMMARY Four main conclusions are drawn from this study.

the use of off-site estimates of PET to estimate wetland ET rates. Understanding the variability of evapotranspiration among wetland communities and throughout the growing season is important for annual or growing season estimates of ET. In the study presented here, there were only 57 days during the 164-day growing season in which all three methods (Penman PET, lysimeter, and water-table method) were available for comparison at both sites (Table 1). Assuming that the average daily values calculated for the early-middle-late period of the growing season are representative of the entire period, a cumulative growing season ET estimate can be made (Table 2). If simple PET rates had been used to estimate the ET loss in a natural wetland, the ET loss calculated by the water-table and lysimeter methods would have been underestimated by 27 and 33 percent, respectively. If water-table and lysimeter measurements from the natural wetland were averaged (growing season ET ⫽ 69.1 cm) and applied to the constructed wetland, the constructed wetland ET loss would have been larger than that calculated by watertable (34 percent over-predicted), lysimeter (49 percent over-predicted), and Penman (40 percent over-predicted) methods. If the period of interest includes a majority of time late in the growing season, PET estimates might also over-predict actual ET rates in the constructed wetland because earlier senescence reduced the wetland’s ability to convey water such that it did not reach the theoretical PET values. While this analysis uses direct measurements as the basis for the comparison, it is clear that appreciable differences in ET rates can be estimated depending on the technique used. Although it is commonly pointed out how hydrology controls the formation of the wetland plant community, the type of plant community became an important forcing function on the hydrology in this study. As a result, it follows that knowledge of the timing and magnitude of ET can be important for designing

(1) A simple meteorological method developed for agricultural applications underestimated the amount of ET in the natural wetland over the majority of the growing season due presumably to violations of limiting assumptions. We believe that this is primarily a result of the surface heterogeneity of the boundary layer resulting in enhanced transport of water vapor from the plant canopy to the unsaturated atmosphere. (2) ET rates were spatially variable, with higher rates measured in the natural wetland. This difference occurred even though the sites were only 300 m apart and were under similar meteorological conditions. This demonstrates the potential problems of extrapolating estimates from a single location over a larger wetland complex and suggests that caution should be used when using off-site estimates of ET. (3) ET rates can be temporally complex in wetlands because they are influenced by the relation of the water table to the root zone and the timing of plant senescence. The relation of ET to water level in the root zone also points out the need to re-evaluate the commonly held assumption that water is seldom limiting in wetlands and water would be removed at the potential evapotranspiration rate. (4) Finally, small-scale geochemical sampling of the shallow root zone was able to elucidate qualitative differences in ET rates. The use of this independent measure allowed identification of higher ET rates in the natural wetlands and differences in the timing of senescence. The methods were not able to estimate ET rates quantitatively, however, due to confounding factors stemming from the threeweek time of close interval membrane equilibrator emplacement. Nor would they be of use when there is not sufficient separation in the geochemical signatures of source end members. Although it is widely recognized that hydrology

Lott & Hunt, ESTIMATING EVAPOTRANSPIRATION controls wetland plant community formation and persistence, in this study, the type of plant community became an important forcing function on the hydrology. It is clear that wetlands can be dynamic systems where abiotic and biotic interaction is all-important, and neither system can be fully understood when decoupled. ACKNOWLEDGMENTS We thank John O. Jackson and John Norman for helpful discussions and Carol Kendall and Steve Silva for lab assistance. The manuscript was improved by critical review from Phil Gerla, Tom Winter, Mike Focazio, and two anonymous reviewers. This work was funded through the University of Wisconsin Systems and the UW-Madison Water Resources Institute. Additional funding was provided by the Wisconsin Department of Transportation and the U.S. Geological Survey. LITERATURE CITED Abtew, W. 1996. Evapotranspiration measurements and modeling for three wetland systems in south Florida. Water Resources Bulletin 32:465–473. Abtew, W. and J. Obeysekera. 1995. Lysimeter study of evapotranspiration of cattails and comparison of three estimation methods. Transactions of the American Society of Agricultural Engineers 38:121–129. Armstrong, J. and W. Armstrong. 1991. A convective throughflow of gases in Phragmites australis. Aquatic Botany 39:75–88. Armstrong, J., W. Armstrong, P. M. Beckett, J. E. Halder, S. Lythe, R. Holt, and A. Sinclair. 1996. Pathways of aeration and the mechanisms and beneficial effects of humidity- and Venturi-induced convections in Phragmites australis. Aquatic Botany 54:177–197. Bravo, H. and G. H. Brown. 1998. 3-D Modeling of groundwater hydrology in a wetland. Advances in Environmental Research 2: 153–166. Carter, V. 1986. An overview of the hydrologic concerns related to wetlands in the United States, Canadian Journal of Botany 64: 364–374. Campbell, D. I. and J. L. Williamson. 1997. Evaporation from a raised peat bog. Journal of Hydrology 193:142–160. Campbell, G. S. and J. M. Norman. 2000. An Introduction to Environmental Biophysics, 2nd ed. Springer-Verlag, Heidelburg, Germany. Chanton, J. P. and G. J. Whiting. 1996. Methane stable isotope distributions as indicators of gas transport mechanisms in emergent aquatic plants. Aquatic Botany 54:227–236. Conservation Foundation. 1988. Protecting America’s Wetlands: an Action Agenda; the Final Report of the National Wetlands Policy Forum. Conservation Foundation, Washington, DC, USA. Dacey, J. W. H. 1981a. Pressurized ventilation in the yellow waterlily. Ecology 62:1137–1147. Dacey, J. W. H. 1981b. How aquatic plants ventilate. Oceanus 24: 43–51. Dooge, J. 1972. The water balance of bogs and fens. p. 233–271. In Proceedings of the Minsk Symposium. UNESCO Press, Paris, France. Dunne, T. and L. B. Leopold. 1978. Water in Environmental Planning. W. H. Freeman and Company, New York, NY, USA. Epstein, S. and T. Mayeda. 1953. Variation of 18O content of water from natural sources. Geochimica Cosmochimica Acta 4:213–244. Gerla, P. 1992. The relationship of water-table changes to the cap-

627 illary fringe, evapotranspiration, and precipitation in intermittent wetlands. Wetlands 12:91–98. Grosse, W., H. Buchel, and H. Tiebel. 1991. Pressurized ventilation in wetland plants. Aquatic Botany 39:98–98. Hammer, D. E. and R. H. Kadlec. 1986. A model for wetland surface water dynamics. Water Resources Research 22:1951–1958. Harbeck, G. E., Jr., M. A. Kohler, and G. E. Koberg. 1958. Waterloss investigations; Lake Mead studies. U.S. Geological Survey Professional Paper 298. Heliotis, F. D. and C. B. DeWitt. 1987. Rapid water table responses to rainfall in a northern peatland ecosystem. Water Resources Bulletin 23:1011–1016. Hesslein, R. H. 1976. An in situ sampler for close interval porewater studies. Limnology and Oceanography 21:912–914. Hughes, P. E., J. S. Hannuksela, and W. J. Danchuck. 1981. Flood of July 15, 1978 on the Kickapoo River, Southwestern Wisconsin. U.S. Geological Survey Hydrologic Investigations Atlas HA 653. Hunt, R. J. 1992. Simulation of drainage ditch and adjacent wetland creation effects on a wetland system using analytic elements. p. 552. In Wetlands: Proceedings of the 13th Annual Conference of the Society of Wetland Scientists, New Orleans, LA. Society of Wetland Scientists, Lawrence, KS, USA. Hunt, R. J., D. P. Krabbenhoft, and M. P. Anderson. 1996. Groundwater inflow measurements in wetland systems. Water Resources Research 32:495–507. Hunt, R. J., D. P. Krabbenhoft, and M. P. Anderson. 1997. Assessing hydrogeochemical heterogeneity in natural and constructed wetlands. Biogeochemistry 34:271–293. Hunt, R. J., T. D. Bullen, D. P. Krabbenhoft, and C. Kendall. 1998. Using stable isotopes of water and strontium to investigate a natural and a constructed wetland. Ground Water 36:434–443. Hunt, R. J., J. F. Walker, and D. P. Krabbenhoft. 1999a. Characterizing hydrology and the importance of ground-water discharge in natural and constructed wetlands. Wetlands 19:458–472. Hunt, R. J., J. O. Jackson, G. L. Running, D. P. Krabbenhoft, and J. T. Krohelski. 1999b. Hydrogeological, geomorphological, and vegetative investigations of select wetland creation and restoration projects. Federal Highways Administration, Washington, DC, USA. Madison, WI: Department of Transportation Technical Report SPR-0092-45-91. Idso, S. B. 1981. Relative rates of evaporative water losses from open and vegetation covered water bodies. Water Resources Bulletin 17:46–48. Idso, S. B. and M. G. Anderson. 1988. A comparison of two recent studies of transpirational water loss from emergent aquatic macrophytes. Aquatic Botany 31:191–195. Kadlec, R. H. and R. L. Knight. 1996. Treatment Wetlands. CRC Press, Boca Raton, FL, USA. Koerselman, W. and B. Beltman. 1988. Evapotranspiration from fens in relation to Penman’s potential free water evaporation (E) and pan evaporation. Aquatic Botany 31:307–320. Krabbenhoft, D. P., C. J. Bowser, M. P. Anderson, and J. W. Valley. 1990. Estimating groundwater exchange with lakes, 1, The stable isotope method. Water Resources Research 26:2445–2453. Kusler, J. A. and M. E. Kentula. 1989. Wetland Creation and Restoration: the Status of the Science, vol.1- Regional Reviews. U.S. Environmental Protection Agency, Washington, DC, USA. Report 600/3–89/038a. LaBaugh, J. W. 1986. Wetland system studies from a hydrologic perspective. Water Resources Bulletin 22:1–10. Lafleur, M. and W. R. Rouse. 1988. The influence of surface cover and climate on energy partitioning and evaporation in a subarctic wetland. Boundary Layer Meteorology 44:327–347. Lafleur, P. M. 1990. Evapotranspiration from sedge-dominated wetland surfaces. Aquatic Botany 37:341–353. Lafleur, P. M. and N. T. Roulet. 1992. A comparison of evaporation rates from two fens of the Hudson Bay Lowland. Aquatic Botany 44:59–69. Linsley, R. K., M. A. Kohler, and J. L. H. Paulhus. 1982. Hydrology for Engineers. McGraw-Hill Company, New York, NY, USA. Lott, R. B. 1997. Estimating evapotranspiration in natural and constructed wetlands: traditional and geochemical approaches. M.S. Thesis. University of Wisconsin-Madison, Madison, WI, USA.

628 McGuinness, J. L. and E. F. Bordne. 1972. A comparison of lysimeter-derived potential evapotranspiration with computed values. Agricultural Research Service, U.S. Deptartment of Agriculture, Washington, DC, USA. Technical Bulletin 1452. Mitsch, W. J. and J. G. Gosselink. 1993. Wetlands. Van Nostrand Reinhold Company, New York, NY, USA. National Cooperative Highway Research Program. 1996. Guidelines for the Development of Wetland Replacement Areas. National Academy Press, Washington, DC, USA. NCHRP Report 379. National Research Council. 1995. Wetlands: Characteristics and Boundaries, National Academy Press, Washington, DC, USA. O’Brien, A. L. 1982. Rapid water table rise. Water Resources Bulletin 18:713–715. Owen, C. R. 1993. Policy-relevant science: hydrologic functions of an urban wetland. Ph.D. Dissertation. University of WisconsinMadison, Madison, WI, USA. Owen, C. R. 1995. Water budget and flow patterns in an urban wetland. Journal of Hydrology 169:171–187. Price, J. S. 1994. Evapotranspiration from a lakeshore Typha marsh on Lake Ontario. Aquatic Botany 48:261–272. Rushton, B. 1996. Hydrologic budget for a freshwater marsh in Florida. Water Resources Bulletin 32:13–21. Scheffe, R. D. 1978. Estimation and prediction of summer evapo-

WETLANDS, Volume 21, No. 4, 2001 transpiration from a northern wetland. Master’s Thesis. University of Michigan, Ann Arbor, MI, USA. Souch, C., C. S. B. Grimmond, and C. Wolfe. 1998. Evapotranspiration rates from wetlands with different disturbance histories. Wetlands 18:216–229. Stauffer, R. E. 1985. Use of solute tracers released by weathering to estimate groundwater inflow seepage to lakes. Environmental Science and Technology 19:405–411. Sturrock, A. M., T. C. Winter, and D. O. Rosenberry. 1992. Energy budget evaporation from Williams Lake: a closed lake in north central Minnesota. Water Resources Research 28:1605–1617. Thompson, M. A., D. I. Campbell, and R. Spronken-Smith. 1999. Evaporation from natural and modified raised peat bogs in New Zealand. Agricultural and Forest Meteorology 95:85–98. Todd, D. K. 1964. Groundwater. p. 13.1–13.55. In V. T. Chow (ed.) Handbook of Applied Hydrology. McGraw-Hill, New York, NY, USA. Winter, T. C. 1981. Uncertainties in estimating the water balance of lakes. Water Resources Bulletin 17:82–115. Yin, Z. Y. and G. A. Brook. 1992. Evapotranspiration in the Okefenokee Swamp watershed: a comparison of temperature-based and water balance methods. Journal of Hydrology 131:293–312. Manuscript received 29 January 2001; revisions received 30 July 2001; accepted 20 August 2001.