evapotranspiration estimates for greenhouse lettuce using an ...

2 downloads 0 Views 311KB Size Report
ABSTRACT. Lettuce production using the nutrient film technique (NFT) in Brazil has increased considerably since 1990. However, the procedure used to supply ...
EVAPOTRANSPIRATION ESTIMATES FOR GREENHOUSE LETTUCE USING AN INTERMITTENT NUTRIENT FILM TECHNIQUE S. Zolnier, G. B. Lyra, R. S. Gates

ABSTRACT. Lettuce production using the nutrient film technique (NFT) in Brazil has increased considerably since 1990. However, the procedure used to supply the nutrient solution is based on interval timers, which does not account for the dynamic environment. In general, nutrient solution is recirculated for 10 to 15 min followed by a resting period of 15 to 20 min, which is used for saving electricity and to provide oxygen to the roots. The objective of this work was to evaluate the Penman -Monteith model for estimating crop evapotranspiration (ETc ) for three cultivars of lettuce (Grand Rapids, Great Lakes, and Regina). Measurements and evapotranspiration estimates were carried out for 30 min intervals. A closed-loop hydroponic system was specially designed for this research so that measurements of ETc could be conducted for short time periods. The hydroponic system was installed in a greenhouse without environmental controls, as is typical for hydroponic lettuce production in Brazil. Measurements of incident solar radiation, relative humidity, air velocity, air temperature, and nutrient solution temperature were stored every minute. The Penman-Monteith model provided excellent estimates of evapotranspiration for leaf area indices (LAI) greater than 0.5. However, the method was not adequate to estimate ETc in the first two weeks after transplanting (LAI < 0.5). During this period, the model sometimes over- or underestimated ETc , depending on the cultivar. Additionally, the model overestimated ETc early in the morning and late in the afternoon because a constant value of the surface diffusive resistance (rs ) was used. The results of this research can be used to improve the electric energy usage efficiency for hydroponic lettuce production and allow the implementation of automatic systems for adjusting the concentration of nutrients in the irrigation water based on estimates of evapotranspiration. Keywords. Hydroponics, Model-based control, Penman-Monteith.

C

urrently, lettuce is the main crop grown in hydroponic production systems in Brazil, most of which use the nutrient film technique (NFT). The two main reasons for the popularity of lettuce cultivation are easy management and short growth cycle. The nutrient solution is recirculated periodically in the channels, generally for 10 to 15 min followed by a resting period of 15 to 20 min. Thus, irrigation scheduling is fixed during the day and throughout the growing season. This empirical practice of providing water and nutrients is accomplished by interval timers, which are insensitive to the dynamic physical environment. In addition, this method does not account for leaf expansion during the crop cycle, which implies increased water demand. Intermittent timers serve a critical purpose during periods of higher temperatures by supplying oxygen to the roots during the time when nutrient solution is not circulating. This

Article was submitted for review in April 2003; approved for publication by the Structures & Environment Division of ASAE in November 2003. Presented at the 2002 ASAE Annual Meeting as Paper No. 023153. The authors are Sérgio Zolnier, Associate Professor, and Gustavo Bastos Lyra, MS Student, Department of Agricultural Engineering, Federal University of Viçosa, Viçosa -MG, Brazil; and Richard S. Gates, ASAE Member Engineer, Professor and Chair, Department of Biosystems and Agricultural Engineering, University of Kentucky, Lexington, Kentucky. Corresponding author: Rich Gates, Professor and Chair, 128 C. E. Barnhart Building, University of Kentucky, Lexington, KY 40546 -0276; phone: 859-257 -3000; fax: 859-257 -5671; e-mail: gates@ bae.uky.edu.

may not be as critical for environmentally controlled greenhouses with intermittent NFT systems and well designed cooling systems. However, in tropical countries with high radiation levels and temperatures, where a significant number of greenhouses do not have environmental control equipment, the use of a continuous supply of nutrient solution, even using a thin film of solution, may result in lack of oxygen. Under an intermittent NFT system, the plant is subject to increased values of electrical conductivity (EC) as soon as circulation of the nutrient solution is interrupted. The EC increases with time toward a maximum value, which depends on the environment conditions. This is due to the fact that water uptake is larger than the nutrient uptake because of high transpiration. When the nutrient solution is recirculated again, the EC values are reestablished. Of course, EC also changes from the beginning to the end of the hydroponic channel. There is much interest in automatic systems for plant production. Irrigation scheduling in greenhouses can be optimized using growth models that simulate the leaf area expansion with evapotranspiration models that are capable of simulating water demand. This can be done with a short time step throughout the growing season. Not only can the amount of water be estimated, but the appropriate time for activating the irrigation system can also be determined. This technique reduces electricity use on cloudy days and prevents water stress on sunny days. Although several methods have been used for evapotranspiration estimates in open-field conditions, in greenhouses the Penman-Monteith model has been evaluated and used by

Transactions of the ASAE Vol. 47(1): 271-282

E 2004 American Society of Agricultural Engineers ISSN 0001-2351

271

many researchers (Faver and O’Toole, 1989; Zhang and Lemeur, 1992; Bailey et al., 1993; Baille et al., 1994; Seginer, 2002; Zolnier et al., 2000, 2001a, 2001b, 2003). Zolnier (1999) used the Penman-Monteith model for estimating evapotranspiration of poinsettia cuttings (Euphorbia pulcherrima “Freedom Dark Red” Willd. ex Klotzsch) subjected to partly wetted conditions in environmentally controlled chambers. The author reported R2adj values between 0.77 and 0.97, and verified that the model overestimated low crop evapotranspiration (ETc) values, as measured when low radiation values were combined with low air vapor pressure deficit (VPDair) levels. In contrast, the model underestimated ETc when high values of artificial radiation were combined with high levels of VPDair or when plants were subjected to high VPDair under dark conditions. Zhang and Lemeur (1992) carried out an experimental study in greenhouse conditions to estimate evapotranspiration of Ficus benjamina plants with leaf area index (LAI) between 2.0 and 3.4 using the Penman-Monteith model. The surface diffusive resistance term was obtained by dividing the mean stomatal resistance by LAI. Resistance to sensible heat transfer by convection was determined by the classical equations for convective heat transfer. The authors reported that the model slightly overestimated evapotranspiration for the entire range of observed LE values. Another experiment related to evapotranspiration estimates in greenhouses conditions was carried out by Baille et al. (1994), in which evapotranspiration measurements were obtained for nine potted species. Estimated and observed values for hibiscus were in reasonable agreement for the entire range of evapotranspiration, but estimated ETc values were severely overestimated for most measurements on poinsettia. The objective of this research was to evaluate the goodness of fit of estimated values of lettuce ETc by the Penman -Monteith model. The model was evaluated from the time of transplanting, and the diurnal course of evapotranspiration was examined for three cultivars (Grand Rapids, Regina, and Great Lakes). The plants were grown in an intermittent NFT system, which was installed in a greenhouse without environmental controls, as is typical in Brazilian production systems.

MATERIAL AND METHODS EXPERIMENTAL SYSTEM The present work was conducted at the Agricultural Meteorology Experimental Station, which is part of the Agricultural Engineering Department at the Federal University of Viçosa, Viçosa -MG, Brazil. The geographical coordinates are 20° 45′ 45″ S latitude, 42° 52′ 04″ W longitude, and 690 m elevation. The average temperature of the warmest month is 22.3°C, the average temperature of the coldest month is 15.4°C, and the precipitation in the driest month of the winter is 19 mm and in the wettest summer month is 245 mm. According to Köppen Classification, the climate is Cwa (Critchfield, 1974). The experiment was carried out from May 18 to June 9, 2001, in a free-standing greenhouse without environmental controls. The greenhouse structure was a Quonset frame (7 m × 15 m floor area, 3.5 m height) constructed of galvanized structural steel tubing (50 mm dia.). The arches (spaced 2.5 m on center) were supported by 2 m high concrete posts

272

at the sidewalls. The greenhouse was oriented with the ridge running east to west. A white plastic insect screen made from high-density polyethylene (thread and opening sizes of 0.28 and 1.00 mm, respectively) was used on the sidewalls to allow natural ventilation while protecting the plants from insects, enabling pesticide-free production. Lettuce seedlings (cv. Grand Rapids, Great Lakes, and Regina) with three or four fully expanded leaves were transplanted to the NFT growing channels 21 days after sowing, where they remained until the end of the experiment. Greenhouse benches consisted of five channels, each 6.0 m long, with 50 mm holes for plants. This design allowed for the placement of 16 plants m -2, spaced 0.25 m apart on center. The upper surfaces of the channels were coated with a white paint to improve solar radiation reflection and to reduce heating of the root zone. Eight greenhouse benches were used: two for each cultivar, and two for replacing plants used during crop growth analysis. The experiment was carried out as a completely randomized design, with three treatments (lettuce cultivars) and two replications (greenhouse benches), and with benches as the experimental units. Each of the eight greenhouse benches contained five channels with 120 plants. The nutrient solution was applied periodically from 05:00 to 19:00 with pumps activated for 10 min and turned off for 20 min for drainage and to provide oxygen to the rooting system. Additionally, the solution was circulated once at midnight to prevent any possible water stress due to night-time transpiration. The pH and the electrical conductivity (EC) of the nutrient solution were monitored and adjusted daily by injection of acid and stock solutions. EC and pH were maintained at approximately 1.3 dS m -1 and 6.5, respectively. DATA ACQUISITION SYSTEM Environmental data inside the greenhouse were acquired using an automatic data acquisition system operating on a personal computer. A data acquisition board (model CYDAS 1602HR, CyberResearch, Branford, Conn.) with 16 analog channels with 16-bit resolution was installed in the computer’s ISA slot. The environmental sensors were connected to the analog channels. Measurements were made of solar radiation, air velocity (Vair), relative humidity (RH), air temperature (Tair), greenhouse covering temperature, and nutrient solution temperature. The flux density of incident solar radiation was measured with an Eppley pyranometer (model PSP, The Eppley Laboratory, Inc., Newport, R.I.), positioned at 0.50 m above the canopy level (approximately 1.30 m above the greenhouse floor). Tair and RH were measured with a combination probe (model HUM50Y, Vaisala, Woburn, Mass.) enclosed in an aspirated radiation shield. Vair was measured using a hot-wire anemometer (model FMA-903-I, Omega, Stamford, Conn.). Both the combination probe and the hot-wire anemometer were installed at the same level as the pyranometer. Calibrated type-T thermocouples were used for measuring temperatures of both the nutrient solution and the greenhouse covering (polyethylene). Environmental data were logged each minute, resulting in 1440 measurements per variable each day. Each logged value represented an average of five values acquired from the time period between 55 and 59 s of each minute, taken at 1 s intervals. All environmental data were averaged over each ETc measurement period (30 or 60 min).

TRANSACTIONS OF THE ASAE

EVAPOTRANSPIRATION MEASUREMENTS A unique measurement system was designed and constructed to obtain ETc of hydroponic lettuce. This measurement system was connected to the hydraulic system of each greenhouse bench, allowing individual measurements of ETc for each experimental unit (120 plants). The nutrient solution reservoir was hydraulically isolated from the measuring tank during periods of ETc measurements by closing valves. Thus, a control volume was established, so any outflow was evapotranspiration or water and nutrients incorporated into the plant tissue. Evapotranspiration measurements were conducted during five 3-day periods during the crop cycle. The measurement periods were broken by a 2-day period when evapotranspiration was not collected. In this way, ETc data were collected for a total of 15 days during the crop cycle. LAI was measured in the second day of each 3-day period. The ETc measurement system (fig. 1) consisted of vertically mounted PVC pipe (0.72 m high) with transparent polyethylene tubing mounted to its outside and marked with graduations (mm) to measure nutrient solution level. The PVC pipe diameter was 0.15 m for the first two weeks after transplanting and 0.20 m in diameter thereafter. The graduated polyethylene tubing and the PVC pipe are together denoted the measuring tank and were used to provide measurements of the nutrient solution level in the measuring tank at determined time intervals. Measurements were carried out from 08:00 until 16:00 at intervals of 30 or 60 min depending on water demand, which was greatly influenced by LAI and environment conditions. A nutrient solution reservoir with 120 L was periodically connected to the measuring tank to reestablish desired levels of electrical conductivity (EC) and pH. Non-corrected evapotranspiration (ETcn) was obtained by dividing the depleted volume of nutrient solution during a specified time interval by the area of the hydroponic growing bench (7.5 m2). The nutrient solution volume was readily available by multiplying changes in nutrient solution level by the transverse sectional area of the PVC pipe. Because water and nutrients are incorporated into the plant tissue, it was necessary to correct ETcn measurements.

In this research, solution incorporated into the plant tissue was estimated by differences between crop growth rate on fresh and dry weight bases. For a particular ETc measurement period, these rates were estimated using the well-known expolinear crop growth model developed by de Goudriaan and Monteith (1990). The concept of day-degree (thermal time), which was accumulated from time of transplant to a specified crop stage, is the basis of this model. The parameters needed in applying this model were determined previously by Lyra (2001). Therefore, water incorporated into the plant tissue was estimated as follows:  ∂FW ∂DW   DD  Ai =  − ∆t    ∂DD ∂DD   60 N 

(1)

where Ai = water incorporated into the plant tissue (kg m -2) ∂FW = crop growth rate by fresh weight basis ∂DD (kg m -2 °C -1 day -1) ∂DW = crop growth rate by dry weight basis ∂DD (kg m -2 °C -1 day -1) DD = degree-day for a particular day after transplanting (°C day) N = duration of the photoperiod for a particular day (h) ∆t = time interval when ETcn measurements were carried out (h). It is important to notice that ETc correction is not relevant during the initial period after transplanting. Lyra (2001) reported that fresh weight of hydroponic lettuce increased 4.8 (±1.0), 4.7 (±1.2), and 6.6 (±0.1) g m -2 day -1 from 4 to 9 days after transplanting for cultivars Grand Rapids, Regina, and Great Lakes, respectively. However, fresh weight increased 332.7 (±12.7), 356.7 (±27.2), and 320.8 (±3.1) g m -2 day -1 from 29 to 34 days after transplanting, respectively, for the same cultivars. Therefore, these results suggest that ETc correction is recommended after canopy closure because daily changes in fresh weight are relevant, especially for fast-growing crops such as lettuce. Corrected crop evapotranspiration can readily be found by: ETc = ETcn - Ai

(2)

CROP EVAPOTRANSPIRATION ESTIMATES Evapotranspiration inside the greenhouse was estimated using the Penman-Monteith model, whose theoretical development is presented by Monteith (1965). Evapotranspiration, expressed as latent heat flux, can be calculated by the following equation: LE =

Figure 1. Layout of the evapotranspiration measurement system. End view (not to scale).

Vol. 47(1): 271-282

∆ (R n − F) + ρair cpair VPDair /rh ∆ + γ (1 + rs /rh )

(3)

where LE = latent heat flux, ETc hfg (W m -2) (Rn - F) = available energy at the crop surface (W m -2) Rn = net radiation at the crop surface (W m -2) F = sensible heat flux into the nutrient solution, normalized for the area of the growing bench (W m -2)

273



≅ derivative of the saturation vapor pressure with respect to temperature, evaluated at the air dry bulb temperature (Pa °C -1) ρair = density of air (kg m -3) rh = resistance to sensible heat transfer by convection for the whole canopy (s m -1) cpair = specific heat of air at constant pressure (J kg -1 °C -1) γ = psychrometric constant (Pa °C -1) VPDair = air vapor pressure deficit (Pa) rs = canopy surface resistance (s m -1) hfg = latent heat of vaporization (J kg -1). The convective resistance (rh) is generally estimated according to the type of heat transfer regime that dominates the greenhouse environment. In this study, forced convection was the dominant heat transfer mode, and rh was estimated according to Chapman (1984) and Incropera and Dewitt (1996): 1

1

 L 2 6   Pr α u∞   rh = 1.328 LAI

(4)

where α = thermal diffusivity (m2 s -1) L = characteristic dimension, (length + width)/2, (m) LAI = leaf area index (dimensionless) u∞ = air velocity in the greenhouse (m s- 1) Pr = Prandtl number (dimensionless). To estimate the effect of leaf expansion on evapotranspiration during the crop cycle, the canopy surface resistance to the diffusion process of water vapor was parameterized according to changes of LAI by the following relationship: rs = rsr where rs

LAIr LAI

(5)

= virtual canopy surface resistance associated with a given LAI (s m -1) rsr = reference canopy surface resistance associated with the LAIr (s m -1) LAI = current LAI estimated by growth models (dimensionless) LAIr = reference LAI, measured after canopy closure but before head formation (dimensionless). Under similar environmental conditions, the virtual canopy surface resistance will account for changes in evapotranspiration caused exclusively by reduced or increased values of LAI during the growing season as compared to the reference LAI value. It is important to recognize that this procedure eliminates the need for applying a crop coefficient (Kc), which is normally used for estimating ETc under field conditions. Values of rsr and LAIr as well as parameters of the expolinear crop growth model for estimating the current L and LAI were previously determined by Lyra (2001) in an independent experiment using the same lettuce cultivars. The rsr values were determined from inversion of the PenmanMonteith model. For the diurnal period from 08:00 to 16:00, the effects of solar radiation and air vapor pressure deficit on rsr were not statistically significant. Thus, all values of rsr

274

were averaged for each cultivar. Average measured values of rsr were 88 (±1), 105 (±14), and 103 (±11) s m -1, which were associated with LAIr values of 2.40 (±0.02), 3.91 (±0.49), and 2.87 (±0.06) for cultivars Grand Rapids, Regina, and Great Lakes, respectively. In this study, the variation of the nutrient solution temperature in a given time interval was used for calculating the heat flux stored/delivered by the solution. The sensible heat flux into the nutrient solution (F) was obtained as: Qs =

m cps ∆T

(6)

A b ∆t

where Qs = sensible heat flux into the nutrient solution, normalized for the area of the hydroponic growing bench (W m -2) m = mass of the nutrient solution that was circulating in the hydroponic system (kg) cps = specific heat of the nutrient solution, assumed to be equal to the pure water (J kg -1 °C -1) ∆T = temperature variation of the nutrient solution during the ETc measuring period (°C) Ab = area of the hydroponic growth bench (7.5 m2) ∆t = time interval corresponding to the ETc measuring period (s). The procedure adopted for estimating net radiation under greenhouse conditions was similar to that presented by Walker et al. (1983). Shortwave radiation was calculated based on incident radiation at the crop level as measured by the Eppley pyranometer, using a reflection coefficient (albedo) of 0.27 (Frisina and Escobedo, 1998). Longwave radiation was calculated based on the Stephan-Boltzmann equation:

(

L w = ε s τt σ Ti4 − ε a To4

)

(7)

where Lw = longwave radiation exchange (W m -2) εs = average emissivity of the interior surfaces (dimensionless) τt = thermal transmittance (dimensionless) σ = Stefan-Boltzmann’s constant (5.6697 × 10 -8 W m -2 K -4) Ti = greenhouse temperature (K) To = outside temperature (K) εa = apparent atmospheric emissivity (dimensionless). The parameters εs , τt , and εa were assumed to be 0.85, 0.80, and 0.86, respectively (Walker et al., 1983).

RESULTS AND DISCUSSION EVALUATION OF THE PENMAN -MONTEITH MODEL Figures 2 through 4 show experimental values of the latent heat flux (LE) and the corresponding estimated values obtained by the Penman-Monteith model for cultivars Grand Rapids, Regina, and Great Lakes, respectively. The results are presented according to the increased values of LAI encountered during crop growth. Adjusted coefficients of determination (R2adj) between predicted and measured LE ranged from 0.77 to 0.92 for cultivar Grand Rapids, from 0.81 to 0.93 for cultivar Regina, and from 0.73 to 0.88 for cultivar Great Lakes. These values

TRANSACTIONS OF THE ASAE

Figure 2. Observed values of lettuce evapotranspiration expressed as latent heat fluxes and the corresponding estimated values by the Penman-Monteith model. Results refer to lettuce cultivar Grand Rapids for mean LAI of (A) 0.14, (B) 0.28, (C) 0.60, (D) 1.28, and (E) 2.85.

indicate a good correlation among observed and estimated values of ETc. It is important to note that these R2adj values are only important for showing the degree of scattering of the experimental values. This work was compared to others who used the Penman-Monteith model for estimating ETc in greenhouse conditions. The values presented here are similar to those reported by Baille et al. (1994), ranging from 0.29 for poinsettia to approximately 0.80 for most greenhouse potted plants. In contrast, the values are below those mentioned by

Vol. 47(1): 271-282

Zhang and Lemeur (1992) for Ficus benjamina, which ranged from 0.97 to 0.98. The latter authors evaluated three different equations for calculating rh according to the heat transfer mode in greenhouse conditions: free, forced, and mixed convection. The equation used for estimating rh under forced convection was identical to the one used in the present work. Regardless of cultivar, measured and predicted values of ETc are in good agreement for LAI greater than 0.5 (figs. 2

275

Figure 3. Observed values of lettuce evapotranspiration expressed as latent heat fluxes and the corresponding estimated values by the Penman-Monteith model. Results refer to lettuce cultivar Regina for mean LAI of (A) 0.11, (B) 0.27, (C) 0.71, (D) 2.25, and (E) 4.06.

through 4, graphs C, D, and E). However, in general, the model was not adequate to estimate ETc for approximately the first two weeks after transplanting (figs. 2 through 4, graphs A and B). During this period, LAI was less than 0.5 for all cultivars. The Penman-Monteith model overestimated LE for cultivars Grand Rapids and Great Lakes but, in general, underestimated LE for cultivar Regina. Specifically, for the latter cultivar, LE values lower than 7.5 W m -2 were overestimated, whereas LE was underestimated for cultivars Grand Rapids and Great Lakes.

276

For a full crop canopy (figs. 2 through 4, graph E), the model overestimated low values of LE, which are typically measured at the beginning and end of the diurnal period. At mid-day hours, high values of LE were slightly underestimated by the model. In this particular crop stage, LAI values were 2.85, 4.06, and 2.52 for cultivars Grand Rapids, Regina, and Great Lakes, respectively. Two different aspects need to be addressed based on the results presented. The first refers to the lack of accuracy due to extrapolation of rs values for the initial period after

TRANSACTIONS OF THE ASAE

Figure 4. Observed values of lettuce evapotranspiration expressed as latent heat fluxes and the corresponding estimated values by the Penman-Monteith model. Results refer to lettuce cultivar Great Lakes for mean LAI of (A) 0.12, (B) 0.24, (C) 0.55, (D) 1.39, and (E) 2.52.

transplanting when canopy closure was not established, and the second refers to the lack of accuracy for estimating ETc at the beginning and end of the diurnal period. The results imply that it is necessary to improve the accuracy of rs estimates for a non-closed crop canopy, in this case up to two weeks after transplanting for the NFT system and cultivars used. Deviation of the estimated ETc values from the measurements are inherent to the method used for estimating rs during crop growth, which are based on changes of the leaf area (eq. 5). Probably an even more important issue

Vol. 47(1): 271-282

regarding ETc estimation is related to difficulties in estimating net radiation for sparse crops. However, for the worstcase scenario considering all cultivars, estimated errors were less than 20 W m -2. This corresponds to evapotranspiration values of 0.030 mm h -1 (1.88 g plant -1 h -1). Regarding ETc errors at the beginning and end of the diurnal period, this can be corrected if the effects of solar radiation and VPDair on the canopy surface resistance are considered. Jolliet and Bailey (1992) stated that using

277

Figure 5. Diurnal changes of solar radiation (Ri), air vapor pressure deficit (VPDair), evapotranspiration expressed as latent heat flux (LE), and the corresponding estimated LE values by the Penman-Monteith model. Measurements were carried out from 8:00 a.m. until 4:00 p.m. in a greenhouse without environment controls on (A) June 3 (partly cloudy) and (B) June 8 (sunny). Evapotranspiration measurements were performed on lettuce cultivar Grand Rapids with LAI of (A) 1.52 and (B) 3.06.

constant rs for estimating evapotranspiration affects prediction accuracy. In this particular study, evapotranspiration measurements were conducted in a closed hydroponic system in which the nutrient solution was provided periodically during a 10 min period followed by 20 min interruption for increasing O2 levels in the rooting system. In spite of providing satisfactory results for estimating ETc, as shown in figures 2 through 4,

278

this measuring system could be improved. One difficulty associated with measuring ETc in this research included water stored in the rooting system, which becomes more significant with crop growth because it becomes more difficult to completely drain the nutrient solution into the measuring tank. In spite of using 120 plants for each greenhouse bench, the volume of the nutrient solution stored in the rooting system was not identical during all draining

TRANSACTIONS OF THE ASAE

Figure 6. Diurnal changes of solar radiation (Ri), air vapor pressure deficit (VPDair), evapotranspiration expressed as latent heat flux (LE), and the corresponding estimated LE values by the Penman-Monteith model. Measurements were carried out from 8:00 a.m. until 4:00 p.m. in a greenhouse without environment controls on (A) June 3 (partly cloudy) and (B) June 8 (sunny). Evapotranspiration measurements were performed on lettuce cultivar Regina with LAI of (A) 2.14/2.37 and (B) 3.62/4.49.

periods of 20 min, and an inevitable transient effect influenced the precision of ETc measurements. Therefore, the effect of solar radiation and air vapor pressure deficit were not statistically significant in a previous experiment, and in this way a constant rs value was used for estimating of ETc during the diurnal period.

Vol. 47(1): 271-282

These results suggest the need for improving the ETc measuring system such that the water is circulated continuously to eliminate this transient draining period. Evidently, oxygen would have to be applied by an automatic system with continuous monitoring of the dissolved O2 levels. This design would allow measurements of water consumption for periods less than the 30 min, as used in this research, and would improve the precision of the evapotranspiration measurements. 279

Figure 7. Diurnal changes of solar radiation (Ri), air vapor pressure deficit (VPDair), evapotranspiration expressed as latent heat flux (LE), and the corresponding estimated values by the Penman-Monteith model. Measurements were carried out from 8:00 a.m. until 4:00 p.m. in a greenhouse without environment controls on (A) June 3 (partly cloudy) and (B) June 8 (sunny). Evapotranspiration measurements were performed on lettuce cultivar Great Lakes with LAI of (A) 1.33/1.46 (A) and (B) both approximately 2.52.

DIURNAL COURSE OF EVAPOTRANSPIRATION Figures 5 through 7 illustrate diurnal variations of evapotranspiration, expressed as LE (W m -2), as affected by incident radiation and VPDair, for measurements carried out on June 3 and 8 (after canopy closure). LE estimated by the Penman -Monteith model followed measured values according to changes of solar radiation and air vapor pressure deficit.

280

Environmental measurements inside the greenhouse suggest that a time lag exists between maximum values of Ri and VPDair on these two particular days. For partly cloudy conditions (graph A), the maximum air vapor pressure deficit was 1400 Pa, which was observed at approximately 12:30 p.m. At that time, solar radiation was almost 450 W m -2. When the maximum solar radiation was observed (10:30 a.m.), VPDair was about 1000 Pa. For a sunny day (graph B), the maximum VPDair value was 1700 Pa, observed

TRANSACTIONS OF THE ASAE

at approximately 3:00 p.m. However, the maximum solar radiation was observed at noon (approximately 600 W m -2). In addition, for clear sky days (figs. 5 through 7, graph B), solar radiation was symmetrically related to chronological time. In contrast, VPDair was clearly skewed toward post-midday hours and dropped sharply at the end of afternoon. Due to this behavior in greenhouse conditions, it was evident that the nutrient solution should be applied at different time intervals according to the crop evapotranspiration. Currently, NFT systems installed in Brazil for lettuce production use interval timers with identical irrigation scheduling during the daylight period. The development of an automatic system, based on estimated values of crop evapotranspiration, is extremely important for optimization of this technique for plant production. In general terms, predicted values of evapotranspiration were in good agreement with measurements. As discussed previously, the model slightly overestimated ETc values early in the morning (08:00 to 10:00) and provided good estimates thereafter. However, errors associated with ETc estimates early in the morning were below 20 W m -2 (0.03 mm h -1, approximately). In spite of having a larger LAI than cultivars Grand Rapid and Great Lakes, cultivar Regina did not show more evapotranspiration. After canopy closure (figs. 5 through 7, graph B), the approximate maximum measured values of LE were 325 W m -2 for cultivar Grand Rapids and 275 W m -2 for both cultivars Regina and Great Lakes. The higher evapotranspiration values for Grand Rapids and similar results for Regina and Great Lakes can be explained by the canopy surface resistance to evapotranspiration. The Penman-Monteith model was appropriate for simulating the diurnal course of evapotranspiration, expressed as latent heat flux, not only on a partly cloudy day (figs. 5 through 7, graph A) but also with high values of solar radiation (figs. 5 through 7, graph B). The LE curve follows incident radiation, depicting the strong effect of the radiation term in the Penman-Monteith equation and pointing out a lower influence of the aerodynamic term, as represented by VPDair. As mentioned previously, LE was overestimated both early in the morning and late in the afternoon, with less significant errors in the afternoon when VPDair reached higher values, showing that this trend was not influenced by cultivar and does not depend on LAI after canopy closure. For Ri values above 300 W m -2 and VPDair above 1000 Pa, the curve of the estimated ETc values was close to the measured points. Results presented in this article are in agreement with the discussion presented by Allen et al. (1997) regarding the effects of Ri and VPDair on ETc after canopy closure. According to this discussion, under humid and calm conditions, which obviously occur in a non-mechanically ventilated greenhouse, as typically used in Brazil, the combined effect of VPDair and the resistance to sensible heat transfer by convection (rh) on evapotranspiration is less important than solar radiation. In this case, as discussed by the authors, evapotranspiration differences between a reference short grass crop and other agricultural crops do not exceed 5%.

Vol. 47(1): 271-282

SUMMARY AND CONCLUSIONS The present work evaluated the goodness of fit between predicted values of evapotranspiration using the PenmanMonteith model and experimental measurements in a hydroponic system (NFT) installed in a non-environmental ly controlled greenhouse, as is typical in Brazil. Comparisons between experimental and estimated values were carried out during daylight conditions from 08:00 to 16:00 for different periods during the crop growth cycle. It was verified that the Penman-Monteith model accurately estimated ETc for LAI values above 0.5. However, the results were not consistent shortly after transplanting (LAI < 0.5), and the model sometimes overestimated or underestimated ET depending on cultivar. Estimated errors of ETc, inherent to the technique used in this study, are caused by extrapolation of the crop resistance according to changes in the LAI. However, the magnitude of the estimated errors was less than 20 W m -2, which corresponds to 0.03 mm h -1 (1.88 g plant -1 h -1). Additionally, the model tended to overestimate ETc values early in the morning and late afternoon, probably due to the use of an average daily canopy surface resistance, which neglects the effect of the environmental factors on rs. The diurnal course of ETc followed solar radiation, indicating that, for a non-environmentally controlled greenhouse, the influence of Ri on ETc is much more important than VPDair, in particular during calm conditions when low values of air velocity are associated with low VPDair in the range from 1.0 to 1.5 kPa. This suggests that estimates of the net radiation will tremendously affect the prediction of evapotranspiration. Results of this research are important to determine the appropriate interval for applying the nutrient solution under a particular environment condition and for a specific crop growth stage.

REFERENCES Allen, R. G., M. Smith, L. S. Pereira, and W. O. Pruitt. 1997. Proposed revision to the FAO procedure for estimating crop water requirements. Proc. 2nd Int. Symp. on Irrigation of Hort. Crops. K. S. Chartzoulakis, ed. Acta Horticulturae 449(1): 17-33. Baille, M., A. Baille, J. C. Laury. 1994. Canopy surface resistances to water vapour transfer for nine greenhouse pot plant crops. Scientia Horticulturae 57: 143-155. Bailey, B. J., J. I. Montero, C. Biel, D. J. Wilkinson, A. Anton, and O. Jolliet. 1993. Transpiration of Ficus benjamina: Comparison of measurements with predictions of the Penman-Monteith model and a simplified version. Agric. and Forest Meteorology 65(3-4): 229-243. Chapman, A. J. 1984. Heat Transfer. 4th ed. New York, N.Y.: Macmillan. Critchfield, H. J. 1974. General Climatology. 3rd ed. Englewood Cliffs, N.J.: Prentice-Hall. Faver, K. L., and J. C. O’Toole. 1989. Short-term estimation of sorghum evapotranspiration from canopy temperature. Agric. and Forest Meteorology 48(1-2): 175-183. Frisina, V. A., and J. F. Escobedo. 1998. Albedômetro com termopilhas de filmes finos e aplicação na determinação do albedo da cultura da alface (Lactuca sativa, L.) em estufa de polietileno. Revista Brasileira da Agrometeorologia 6(1): 81-90.

281

Goudriaan, J., and J. L. Monteith. 1990. A mathematical function for crop growth based on light interception and leaf area expansion. Annals of Botany 66(6): 695-701. Incropera, F. P., and D. P. DeWitt. 1996. Fundamentals of Heat and Mass Transfer. 4th ed. New York, N.Y.: John Wiley and Sons. Jolliet, O., and B. J. Bailey. 1992. The effect of climate on tomato transpiration in greenhouses: Measurements and models comparison. Agric. and Forest Meteorology 58(1-2): 43-62. Lyra, G. B. 2001. Estimativa da evapotranspiração e análise de crescimento para alface (Lactuca sativa L.) cultivada em sistema hidropônico em condições de casa-de-vegetação. MS thesis (in Portuguese). Viçosa, Brazil: Federal University of Viçosa: Department of Agricultural Engineering. Monteith, J. L. 1965. Evaporation and environment. Symposia of the Society for Experimental Biology 19: 205-234. Seginer, I. 2002. The Penman-Monteith evapotranspiration equation as an element in greenhouse ventilation design. Biosystems Eng. 82(4): 423-439. Walker, J. N., R. A. Aldrich, and T. H. Short. 1983. Chapter 8: Quantity of air flow for greenhouse structures. In Ventilation of Agricultural Structures, 169-191. M. A. Hellickson and J. N. Walker, eds. St. Joseph, Mich.: ASAE.

282

Zhang, L., and R. Lemeur. 1992. Effect of aerodynamic resistance on energy balance and Penman-Monteith estimates of evapotranspiration in greenhouse conditions. Agric. and Forest Meteorology 58(3-4): 209-228. Zolnier, S. 1999. Dynamic misting control techniques for poinsettia propagation. PhD diss. Lexington, Ky.: University of Kentucky, Department of Biosystems and Agricultural Engineering. Zolnier, S., R. S. Gates, J. W. Buxton, and C. Mach. 2000. Psychrometric and ventilation constraints for vapor pressure deficit control. Computers and Electronics in Agric. 26(3): 343-359. Zolnier, S., R. S. Gates, R. G. Anderson, S. E. Nokes, and G. A. Duncan. 2001a. Non-water-stressed baseline as a tool for dynamic control of a misting system for propagation of poinsettias. Trans. ASAE 44(1): 137-147. Zolnier, S., R. S. Gates, R. L. Geneve, and J. W. Buxton. 2001b. Surface diffusive resistances of rooted poinsettia cuttings under controlled-environment conditions. Trans. ASAE 44(6): 1779 -1787. Zolnier, S., R. S. Gates, R. L. Geneve, and J. W. Buxton. 2003. Evapotranspiration-based misting control for poinsettia propagation. Trans. ASAE 46(1): 135-145.

TRANSACTIONS OF THE ASAE