Abstract
Greenhouse energy-saving and biocide reduction can be achieved through dynamic greenhouse climate control with computerized model-based regimes. This can be optimized when next to greenhouse macroclimate (i.e., the aerial environment) also, the crop microclimate is predicted. The aim of this article was to design and apply a simple deterministic microclimate model for dynamic greenhouse climate control concepts. The model calculates crop temperature and latent heat of evaporation in different vertical levels of a dense canopy of potted plants. The model was validated with data attained from experiments on dynamic or nondynamic (regular) controlled greenhouse cultivation. Crop temperature was with a 95% confidence interval of 2 °C or 2.4 °C for sunlit or shaded leaves, respectively, accurately predicted in a simple greenhouse with predefined climate set points. With a more dynamic greenhouse control also including assimilation lighting and screens, the prediction quality decreased but still had a 95% confidence interval of crop temperature prediction of 3.8 °C for sunlit leaves. Simulations showed that controlling greenhouse temperature according to the predicted crop temperature rather than according to the air temperature can save energy. Energy-saving is highest during winter and 12% energy saving was attained during January under Danish climate conditions.
During the last 20 years, several dynamic temperature regimes were designed for greenhouse energy-saving (e.g., Aaslyng et al., 1999; Bailey, 1985; Bailey and Seginer, 1989; Buwalda et al., 1999; Körner and Challa, 2003a; Seginer et al., 1994). In the early 1990s, a complete dynamic climate control concept was first developed (Aaslyng et al., 2003) and constantly further developed since then. The system aims at optimizing the greenhouse microclimate to ensure maximum net dry matter production, taking energy consumption into consideration (Aaslyng et al., 2003).
It is a component-based system, i.e., it consists of separate software building blocks that can be updated separately and new components can be added easily. Each component handles a biologic, physical, or environmental control task. Next to a component for dynamic temperature control, other climatic control components for, e.g., pests (Jakobsen et al., 2005), diseases (Körner and Holst, 2005), or supplementary assimilation light (Aaslyng et al., 2006, Körner et al. 2006), were developed.
Controlling greenhouse climate dynamically results in fluctuating greenhouse air temperature. Greenhouse air temperature and crop temperature influence each other, but due to the different speed of warming up and cooling down, crop temperature can be higher or lower than greenhouse air temperature. As it is the crop temperature that influences crop growth and developmental processes, crop temperature rather than greenhouse air temperature needs to be controlled. In addition, the difference between air and crop temperature and the speed of temperature changes are the keys for many plant diseases (Körner and Challa, 2003b; Körner and Holst, 2005).
Although several approaches were presented to calculate crop temperature from the greenhouse macroclimate (Kempkes and Van de Braak, 2000; Wang and Deltour, 1999; Zhang et al., 1997, 2002), most climate controllers and regimes control the greenhouse air temperature rather than crop temperature. The few approaches to control crop temperature were designed for and tested with nondynamic (i.e., regular) greenhouse temperature regimes only. The few microclimate models for dynamic climate regimes (Boonen et al., 2000; Van Pee et al., 1998) were developed as black box models that are not generically usable. The aim of this work was therefore to develop and apply a simple generic deterministic microclimate model that can be used for dynamic greenhouse climate control. The model should be able to predict crop microclimate from a greenhouse macroclimate (monitored by commercial climate-measuring equipment). In that way, greenhouse microclimate can be predicted by regular sensors that are located above the crop rather than distributing sensors within the crop.
A model is presented that describes the crop microclimate in a vertical transect of the crop canopy. The model was validated with experimental data from dynamic and regular climate regimes. In a final simulation study, comparisons were made on the effects on energy consumption and crop microclimate by controlling greenhouse temperature either based on ambient greenhouse temperature (belonging to macroclimate) or by crop temperature (belonging to the microclimate) calculated with the presented microclimate model.
Materials and Methods
Model development.
Crop temperature (Tc, K) in different canopy layers (z) was calculated by integrating absorbed irradiative net fluxes (Rn,a, W·m−2), boundary layer and stomata resistances (rb and rs, respectively, s·m−1), and vapor pressure deficit at the leaf surface (VPDs, kPa) in the canopy. The extinction of short-wave radiation as it passes through the canopy to heights z was calculated with the Lambert-Beer's law; z was a function of the leaf area index (LAI) and estimated for three vertical positions in the canopy according to the Gaussian three-point integration procedure (Goudriaan and Van Laar, 1994):
Too high potential λE can result in a higher transpiration than plants can handle and then water loss may exceed water uptake (Körner and Challa, 2003b). In the opposite case when λE becomes negative, dew will be formed on leaves and other tissues that can lead to plant diseases.
The penetration function for upwards long-wave radiation, τu, was assumed similar to τd. Tr was either calculated from the four variables greenhouse air temperature, effective sky temperature (Yu et al., 2001), outside temperature, and wind speed outside the greenhouse (De Zwart, 1996); or when available, measured data were used to predict Tr. Tg was either assumed to be the same as greenhouse air temperature or measured data were used.
Absorbed I PPF fluxes were calculated separately for diffuse and direct radiation and as a function of z (Van Kraalingen and Rappoldt, 1989). Leaf photochemical efficiency, αl , and Pnl,max were calculated based on biochemical leaf photosynthesis models (Farquhar and Von Caemmerer, 1982; Farquhar et al., 1980) and approaches of Gijzen (1994) as shown by Körner (2004). The boundary layer resistances (rb, s·m−1) for leaves in different crop levels z were attained from the dimensionless Nusselt number (Nu) and the thermal diffusivity of air (χ, m2·s−1) as proposed for potted chrysanthemums (Yang, 1995). The Nu number was calculated from the dimensionless Grashof (Gr) and Reynolds (Re) numbers (Stanghellini, 1987):
with air velocity (u, m·s−1), the characteristic dimension of the leaf surface (l, m), the kinematic viscosity of air (υ, m2·s−1), acceleration resulting from gravity (g, m·s−1), and the coefficient of thermal expansion (β, K−1).
Greenhouse experiments.
Two experiments with potted roses were carried out. Expt. 1 was conducted with a 24-h dynamic temperature regime, i.e., temperature was only dynamic within 24 h and set points were the same for each 24 h. Expt. 1 was a short-term experiment of 5 weeks (8 Nov.–15 Dec. 2004). Roses (Rosa, ‘Vanilla Charming Parade’) were planted in black plastic pots (8-cm high, 10-cm diameter) filled with a commercial peat mix and were then placed on an ebb and flow table in a research greenhouse compartment (25 m2) at the Danish Institute of Agricultural Sciences (Slagelse, Denmark, 55°18′N). Plants were placed with a spacing of 10 cm to all sides. Plants were allowed to grow naturally (no pinching), but plant height was kept at 20 cm (from pot edge) by periodic cutting with electric scissors. Within each 24-h period, temperature was set to 10, 20, or 24 °C:10 °C (0800–1159 hr; 2000–2359 hr), 20 °C (1200–1759 hr; 0000–0559 hr), and 24 °C (0600–0759 hr; 1800–1959 hr). As a result of the relatively large greenhouse surface in relation to the plant canopy, humidity levels were expected to be much lower than in commercial size greenhouses. To attain humidity levels like in commercial practice, the relative humidity (RH) set point was 100% (i.e., neither heating nor ventilation for dehumidification was used). In addition, a fogging and ventilation system located under the tables distributed water vapor in the greenhouse when RH was lower than 85%. With that procedure, RH levels with an average of 74% (ranging between 52% and 100% RH, standard deviation 15.3% RH) were attained. Only during some periods, RH was close to 100% and no incidences with fungi such as Botrytis were observed before determination of the experiment.
Climate was controlled with a commercial greenhouse climate computer (LCC1200; DGT Volmatic, Odense, Denmark). The vents were closed throughout the complete experiment. Pipe heating was used. An energy-saving screen (LS10; Ludvig Svensson, Kinna, Sweden) was closed during night between 1900 and 0600 hr. No assimilation light was used. Greenhouse macroclimate (RH and air temperature) was measured with a commercial climate measuring box (0.8 m above the crop) equipped with a capacitive hygrometer and a PT500 thermometer (Senmatic, Søndersø, Denmark). Microclimate was measured in the two crop levels z1 and z3. In both levels, leaf surface temperature was measured with two IR thermocouples (IRTS-P5; Apogee Instruments, Logan, UT); RH and temperature close to the leaf surfaces (from 3 to 5 cm distance) were measured with a capacitive hygrometer and a PT100 thermometer, respectively (Hygroclip S3; Rotronic, Basserdorf, Switzerland). Because z1 and z3 were a function of the LAI (Eq. [12]), the measuring equipment was adjusted weekly. PPF density (I PPF , μmol·m−2·s−1) was measured with two line quantum sensors (LI-190SA; LI-COR, Lincoln, Neb.). All data were continuously recorded and logged with 1-min averages in two computer-controlled data loggers (DT 600 with DeLogger 4; DataTaker, Rowville, Australia).
In Expt. 2, temperature was controlled by the IntelliGrow system, a complex control regime (Aaslyng et al., 2003). Expt. 2 compared the microclimate attained with a regular and a dynamic temperature regime. Within the dynamic regime, temperature could fluctuate differently between days; set points were fixed for the regular regime (Table 1). Expt. 2 was a long-term experiment (5 Sept. 2001–27 May 2002) with a total of seven batches of potted roses (Rosa, ‘Vanilla Charming Parade’) grown sequentially until plants had reached the commercial harvest stage. Young potted roses were placed to a density of 40 plants·m2 on ebb and flow tables (7 m2) in four greenhouse compartments (66 m2 each) of the Royal Veterinary and Agricultural University (Copenhagen, Denmark, 55°41′N).
Overview of the set points used in the dynamic climate regime in Expt. 2 (Lund et al., 2006).
In Expt. 2, outdoor climate was measured by a standard weather station. Greenhouse macroclimate as air temperature (10K3MCD1; BetaTHERM, Galway, Ireland), CO2 concentration (from 0 to 2000 μmol·mol−1 ± 10 μmol·mol−1) (URAS 4; Hartmann & Braun, Frankfurt, Germany), air humidity (from 0% to 100% RH, ±1.5%) (Hygroclip S; Rotronic, Basserdorf, Switzerland), global radiation (CM-11 pyranometer; Kipp and Zonen, Delft, The Netherlands), and I PPF (G1125–06; Hammamatsu, Japan; and LI-190SA; LI-COR) were measured and computer-controlled (LCC1200; DGT Volmatic, Odense, Denmark). The climate set points were generated with the IntelliGrow software (Aaslyng et al., 2003) that was implemented in another computer and sent to the LCC1200 in a 10-min time interval (Aaslyng et al., 2005). Set points were determined from macroclimate measurements and either calculated dynamically or by the regular regime (Table 1). Greenhouse microclimate as temperature (four evenly distributed type T thermocouples and one infrared [IR] thermocouple) (Sensycon; Hartmann & Braun, Frankfurt, Germany; Exergen IR, Watertown, Mass., respectively), I PPF (G1125–06; Hammamatsu, Japan; and LI-190SA; LI-COR), and global radiation (CM-3 pyranometer; Kipp and Zonen) were measured on each table on sunlit leaves only (position z1). Greenhouse compartments roof temperatures were measured with thermistors (100K6MCD1; BetaTHERM, Shrewsbury, Mass.) on the north and on the south side of each greenhouse compartment. The same type of thermistors were also used to measure soil temperature and pipe heating in- and outlet temperatures. All data were continuously measured, averaged over 5 min, and stored in a data logger (CR10X; Campbell Scientific, Logan, Utah). Supplementary lighting was used (see Table 1). Ten high-pressure sodium lamps (SON-T, 400 W; Philips, Eindhoven, The Netherlands) were installed in each greenhouse compartment (0.152 lamps per m2). Next to visible radiation, the high-pressure sodium lamps produced 80 W IR radiation and 202 W heat plus an additional 36 W produced by the generator. For simple estimation of long-wave radiation exchange with the upper leaf surface (Eq. [11]), it was assumed that 50% of the heat production reached the crop top surface.
Simulating greenhouse microclimate.
The previously described model was programmed in the simulation software environment MATLAB (version 7.2; The MathWorks; Lowell, Mass.). The model's sensitivity was tested with artificially created climate data (Table 2). Simulations were performed with measured greenhouse macroclimate data (Expts. 1 and 2) using a 5-min simulation time step.
Simulations settings for climate control study with simulation range (range) and simulation time steps (step unit).
Climate control simulation study.
For a simulation study, a greenhouse climate and control simulator (GCCS) (Körner et al., 2004) was connected to the presented microclimate model. The GCCS calculated greenhouse macroclimate for a standard 1-ha Venlo-type greenhouse with a single glass cover (transmission for diffuse radiation of 78.5%), energy-saving screen (LS10; Ludvig Svensson, Kinna, Sweden), and assimilation light. The Danish 1-year reference climate data file (Design Reference Year [DRY]; Lund, 1995) was input to the GCCS. Control inputs were climate set points for heating, ventilation, CO2 concentration, and screening. Those were generated by a set point generator program (SPG) and provided to the GCCS by data file-sharing in a 5-min interval. This simulation study compared the effect on energy consumption and the microclimate factors dew formation and crop temperature of 1) commonly used greenhouse climate control with greenhouse air temperature with 2) greenhouse climate control with crop temperature, meaning the SPG either used greenhouse air temperature or crop temperature as the basis for set point calculations. Two temperature regimes were tested: 1) regular temperature set points for day and night with a 1 °C margin between heating and ventilation, and 2) a dynamic 24-h temperature integration regime in which temperature could fluctuate freely between the calculated heating and ventilation temperatures (Table 3). Both regimes were thoroughly described in Körner et al. (2004).
Climate settings for simulation study for regular and dynamic (24-h temperature integration) climate control: ventilation temperature (Tv), heating temperature (Th), relative humidity (RH), CO2 concentration.
Results
Model presentation.
The model predicted a strong dependency of crop temperature to short-wave radiation and greenhouse air temperature. The effect of short-wave radiation decreased with increasing LAI (i.e., the effect on z1 leaves was higher than on z2 or z3 leaves) and decreased with increasing greenhouse air temperature (Figs. 1 and 2). The difference between crop temperature and cover temperature increased with an increasing gradient between inside and outside greenhouse temperature. In this situation, leaves exchange less long-wave radiation if they are situated in lower canopy levels and therefore lose less energy than leaves in higher levels. This leads to temperature difference between z1 and z3 leaves. Increasing the short-wave radiation had the opposite effect (Figs. 1 and 2). Strong effects of short-wave radiation on potential evaporation could be observed, too. When short-wave radiation was high, λE decreased with increasing LAI (Fig. 3). At darkness, λE reacted opposite to that, and when RH was very high, negative λE values could be observed in z1 leaves, i.e., condensation on the plant surface and an increasing risk of diseases (Fig. 3). The effects of short-wave radiation on both leaf temperature and dew formation (i.e., negative λE) decreased with increasing LAI (Figs. 3 and 4). Those effects were strongest when short-wave radiation was high, because the position of z in the canopy was not constant but defined by LAI (Eq. [1]).
Simulated crop temperature in three canopy positions (z1 [—], z2 [– –], and z3 [—]) as a function of greenhouse air temperature for four inside global radiation levels: (A) 0 W·m−2, (B) 40 W·m−2, (C) 80 W·m−2, and (D) 160 W·m−2. All other factors were set constant: leaf area index 2, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated crop temperature in three canopy positions (z1 [—], z2 [– –], and z3 [—]) as a function of greenhouse air temperature for four inside global radiation levels: (A) 0 W·m−2, (B) 40 W·m−2, (C) 80 W·m−2, and (D) 160 W·m−2. All other factors were set constant: leaf area index 2, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated crop temperature in three canopy positions (z1 [—], z2 [– –], and z3 [—]) as a function of greenhouse air temperature for four inside global radiation levels: (A) 0 W·m−2, (B) 40 W·m−2, (C) 80 W·m−2, and (D) 160 W·m−2. All other factors were set constant: leaf area index 2, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated crop temperature at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at (A) 0 W·m−2 or (B) 160 W·m−2 global radiation inside the greenhouse. All other factors were set constant: LAI 1.5, RH 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated crop temperature at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at (A) 0 W·m−2 or (B) 160 W·m−2 global radiation inside the greenhouse. All other factors were set constant: LAI 1.5, RH 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated crop temperature at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at (A) 0 W·m−2 or (B) 160 W·m−2 global radiation inside the greenhouse. All other factors were set constant: LAI 1.5, RH 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated latent heat of evaporation at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at 0 W·m−2 (gray lines) or 160 W·m−2 (black lines) global radiation inside the greenhouse at (A) 70% or (B) 100% relative humidity. All other factors were set constant: leaf area index 1.5, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated latent heat of evaporation at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at 0 W·m−2 (gray lines) or 160 W·m−2 (black lines) global radiation inside the greenhouse at (A) 70% or (B) 100% relative humidity. All other factors were set constant: leaf area index 1.5, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated latent heat of evaporation at 15 °C (—), 25 °C (– –), and 35 °C (—) air temperature as a function of the canopy level z at 0 W·m−2 (gray lines) or 160 W·m−2 (black lines) global radiation inside the greenhouse at (A) 70% or (B) 100% relative humidity. All other factors were set constant: leaf area index 1.5, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Crop temperature at different inside greenhouse global radiation levels of (A, B) 0 W·m−2 and (C, D) 160 W·m−2 as a function of the canopy level z and at (A, C) four different leaf area indexes (LAI 0.1 [—], LAI 1 [– –], LAI 2 [—], LAI 3 [- – -]) and (B, D) three different canopy levels (z1 [—], z2 [—], and z3 [– –]). All other factors were set constant: greenhouse air temperature 20 °C, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2) LAI indicates leaf area index.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Crop temperature at different inside greenhouse global radiation levels of (A, B) 0 W·m−2 and (C, D) 160 W·m−2 as a function of the canopy level z and at (A, C) four different leaf area indexes (LAI 0.1 [—], LAI 1 [– –], LAI 2 [—], LAI 3 [- – -]) and (B, D) three different canopy levels (z1 [—], z2 [—], and z3 [– –]). All other factors were set constant: greenhouse air temperature 20 °C, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2) LAI indicates leaf area index.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Crop temperature at different inside greenhouse global radiation levels of (A, B) 0 W·m−2 and (C, D) 160 W·m−2 as a function of the canopy level z and at (A, C) four different leaf area indexes (LAI 0.1 [—], LAI 1 [– –], LAI 2 [—], LAI 3 [- – -]) and (B, D) three different canopy levels (z1 [—], z2 [—], and z3 [– –]). All other factors were set constant: greenhouse air temperature 20 °C, relative humidity 80%, outside air temperature 15 °C, CO2 350 μmol·mol−1 (see Table 2) LAI indicates leaf area index.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Greenhouse experiments.
The model was able to predict leaf surface temperature well for both z1 and z3 leaves when temperature fluctuations were preprogrammed (Expt. 1) (Fig. 5). However, periodically, some incongruence occurred. The natural heating up and cooling down of the greenhouse (no heating) was well predicted (high and low peaks). Only when pipe heating was used to heat up the greenhouse after a cool period did the model disagree with the measurements. However, the difference in predicted and measured temperature was noteworthy only during these periods.
Measured (black) and simulated (gray) crop temperature for (A) z3 leaves and (B) z1 leaves during 10 d of Expt. 1 (day 335 to day 344 of the year). The position of the indicated day numbers is at 1800 hr.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Measured (black) and simulated (gray) crop temperature for (A) z3 leaves and (B) z1 leaves during 10 d of Expt. 1 (day 335 to day 344 of the year). The position of the indicated day numbers is at 1800 hr.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Measured (black) and simulated (gray) crop temperature for (A) z3 leaves and (B) z1 leaves during 10 d of Expt. 1 (day 335 to day 344 of the year). The position of the indicated day numbers is at 1800 hr.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
In situations when the greenhouse climate was more actively controlled than in Expt. 1, i.e., also with artificial lighting, screens, and vents (Expt. 2), the model had to be adjusted to these circumstances. When energy exchanges with the lamps and the screen were added to the model, greenhouse microclimate was well predicted (Fig. 6).
Measured (black) and simulated (gray) pot-rose crop temperature with data from Expt. 2 for 40 d (A) (between day of the year 108 and 148) and day of year 147 (B) for a dynamic climate regime with the IntelliGrow system.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Measured (black) and simulated (gray) pot-rose crop temperature with data from Expt. 2 for 40 d (A) (between day of the year 108 and 148) and day of year 147 (B) for a dynamic climate regime with the IntelliGrow system.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Measured (black) and simulated (gray) pot-rose crop temperature with data from Expt. 2 for 40 d (A) (between day of the year 108 and 148) and day of year 147 (B) for a dynamic climate regime with the IntelliGrow system.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
The model's high quality is expressed by the narrow 95% confidence interval in Expt. 1 that was ≈2 °C for z1 leaves and 2.4 °C for z3 leaves (Fig. 7). Simulated and measured crop temperature differed only in some cases by more than 2 °C. In more than 60%, predictions were within a 1 °C range (60% and 66% for z1 and z3, respectively) (Fig. 8). The 95% confidence interval in Expt. 2 was much higher than in Expt. 1 (i.e., 3.8 °C, Fig. 9).
Simulated versus measured leaf temperature for (A) z1 leaves and (B) z3 leaves in Expt. 1 with 95% nonsimultaneous confidence intervals: (A) ±1.99 °C and (B) ±2.43 °C.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated versus measured leaf temperature for (A) z1 leaves and (B) z3 leaves in Expt. 1 with 95% nonsimultaneous confidence intervals: (A) ±1.99 °C and (B) ±2.43 °C.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated versus measured leaf temperature for (A) z1 leaves and (B) z3 leaves in Expt. 1 with 95% nonsimultaneous confidence intervals: (A) ±1.99 °C and (B) ±2.43 °C.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Frequency diagrams of absolute deviations from simulated to measured leaf temperature in Expt. 1 for (A) z1 and (B) z3 and (C) Expt. 2.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Frequency diagrams of absolute deviations from simulated to measured leaf temperature in Expt. 1 for (A) z1 and (B) z3 and (C) Expt. 2.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Frequency diagrams of absolute deviations from simulated to measured leaf temperature in Expt. 1 for (A) z1 and (B) z3 and (C) Expt. 2.
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated versus measured leaf temperature for z1 leaves during 40 d in Expt. 2 with the 95% nonsimultaneous confidence interval (±3.8 °C).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated versus measured leaf temperature for z1 leaves during 40 d in Expt. 2 with the 95% nonsimultaneous confidence interval (±3.8 °C).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Simulated versus measured leaf temperature for z1 leaves during 40 d in Expt. 2 with the 95% nonsimultaneous confidence interval (±3.8 °C).
Citation: HortScience horts 42, 2; 10.21273/HORTSCI.42.2.272
Climate control.
The simulation study with the dynamic climate regime showed a closer fit between microclimate and set points when greenhouse climate was controlled to crop temperature (rather than to greenhouse air temperature) (Table 4). Because results with the regular climate regime were similar to the dynamic regime, no further data are presented. Greenhouse air temperature control resulted in warmer plants than wanted throughout the year. The difference in crop temperature between air and crop temperature control was most eminent in winter. In January, leaf temperature in all three vertical crop levels was ≈1 °C higher with greenhouse air temperature control than with crop temperature control. For that, unnecessary heating energy was used (Table 4). Shifting from greenhouse air temperature control to crop temperature control with the presented model could save more than 12% energy in January. Energy-saving percentages decreased toward summer. Only a small positive effect of crop temperature control on dew formation could be observed (Table 4).
Weekly averages of air temperature, crop temperature, and sums for dew; and energy consumption in three canopy levels z for simulations with dynamic climate control (see Table 3) with air temperature or crop temperature control (ATC or LTC) and standard deviation.z
Discussion
Microclimate prediction models applicable for model-based greenhouse climate control are scarce. We therefore aimed at creating a simple microclimate model with the prediction quality demanded in dynamic greenhouse climate control. Quality assessment of a crop microclimate model quality includes mainly five major points: 1) general fit between real and simulated crop temperature, 2) prediction quality of high and low temperature peaks, 3) predicting the difference in temperature fluctuation speed between air and plants, 4) predicting the vertical microclimate differences in the crop canopy, and 5) prediction of the horizontal microclimate differences in the greenhouse.
The general fit between simulated and measured crop temperature was satisfying. However, the model yielded in some biased predictions of leaf temperature, with the largest over- and underestimations during warming up of the greenhouses through pipe heating. Estimating leaf temperature was however successful when temperature was not controlled. The discrepancies between measured and simulated leaf temperatures in some periods can probably be explained by the complexity of the greenhouses the measurements were taken in. In addition, microclimate predictions with the complete dynamic regime and with more active climate control and assimilation light (Expt. 2) had a lower quality. This could probably be explained through strong simplifications in the model construction for, e.g., artificial assimilation light or heating pipes. Also, greenhouse size, position of the heating pipes, table structures and positions, and table distances from heating pipes, and so on, differ between greenhouses and the greenhouse equipment, which was only approximated and roughly implemented in the model. Because greenhouses in northern Europe are often equipped with assimilation lights, a better implementation of those (and other equipment) will strongly improve the quality of the model and its potential for the use in practice. There are many parameters that could improve the models prediction quality, e.g., heat transfer calculations between heating pipes and tables, tables and pots, and plants and side walls. Other parameters such as the influence of water on the tables on crop microclimate, the thickness of leaves, or the compactness of the plants was not taken into account either. Furthermore, assuming that the sum of all nonliving greenhouse material temperatures (Tg) is the same as greenhouse air temperature may not be correct, but to include the complete interior of the greenhouse involves complex calculations and model adjustments. Different materials can either be warmer or colder than the air (depending on greenhouse climate, material, and humidity content). Because we aimed at developing a simple and generic microclimate prediction model, we judge the simplifications that were made as reasonable for an overall estimation.
With only a few adjustments, the model was able to predict greenhouse microclimate well in different greenhouses such that the 95% confidence interval was 2 °C or 3.8 °C. Because the model was able to predict high and low peaks very well and because this is the most important quality criterion, the incongruence in the general fit are acceptable within its low range. Also, the slow reaction of the crop tissue to temperature fluctuations was well predicted (Fig. 6). The last criterion, however, was not implemented in the model nor was it tested. Calculating the horizontal microclimate differences in greenhouses demands a complex three-dimensional model and this is out of scale of the present research. However, our experiments were done in small research facilities that are very different from commercial-sized greenhouses that can measure several hectares. The climate in large greenhouses is probably more uniform and a general model, like the one presented here, could probably be applied much easier under such conditions. The present model was demonstrated through sensitivity analysis and through model validation with data from two different research greenhouses. Although the accuracy of the predictions differed in the two greenhouse experiments, these differences were small.
The presented microclimate model has the potential for being generic. A simulation case study showed that using this model to predict crop microclimate is a promising alternative to greenhouse air temperature control and may be used for model-based dynamic control regimes (Aaslyng et al., 1999; Bailey, 1985; Bailey and Seginer, 1989; Buwalda et al., 1999; Körner and Challa, 2003a; Seginer et al., 1994) or for the concept of optimal greenhouse climate control (Gal et al., 1984). Because 10% energy is saved by a 1 °C lower greenhouse temperature (Tantau, 1998), heating demand and therefore energy consumption can strongly decrease when using this microclimate model for crop temperature control. This was also shown in the simulation study. Controlling the greenhouse climate more specific to the coldest crop layer (rather than using an average crop temperature as the control) and designing a regime for crop diseases can improve the concept strongly. Condensation could be reduced and therefore also the risk for fungi diseases (Huber and Gillespie, 1992; Körner and Holst, 2005). For the optimal use of that, however, the previously mentioned three-dimensional microclimate prediction is needed.
It can thus be concluded that the presented model is useful for microclimate predictions, although its robustness remains to be tested for additional greenhouse constructions and locations and different plant species. Strong improvements to the model would be a three-dimensional greenhouse macroclimate model that could be used to predict horizontal greenhouse microclimate differences.
Notation
Literature Cited
Aaslyng, J.M. , Ehler, N. & Jakobsen, L. 2005 Climate control software integration with a greenhouse environmental control computer Environ. Model. Softw. 20 521 527
Aaslyng, J.M. , Ehler, N. , Karlsen, P. & Rosenqvist, E. 1999 Intelligrow: A component based greenhouse climate control system for decreasing energy consumption Acta Hort. 507 35 41
Aaslyng, J.M. , Körner, O. , Andreassen, A.U. , Lund, J.B. , Skov, J. , Ottosen, C.O. & Rosenqvist, E. 2006 Integrated optimization of temperature, CO2, screen and artificial lighting in greenhouse crops Acta Hort. 711 79 88
Aaslyng, J.M. , Lund, J.B. , Ehler, N. & Rosenqvist, E. 2003 Intelligrow: A greenhouse component-based climate system Environ. Model. Softw. 18 657 666
Bailey, B.J. 1985 Wind dependent control of greenhouse temperature Acta Hort. 174 381 386
Bailey, B.J. & Seginer, I. 1989 Optimum control of greenhouse heating Acta Hort. 245 512 518
Ball, J.T. , Woodrow, I.E. & Berry, J.A. 1987 A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions 221 224 Biggins J. Progress in photosynthesis research Martinus Nijhoff Dordrecht, The Netherlands
Boonen, C. , Joniaux, O. , Janssen, K. , Berckmans, D. , Lemeur, R. , Kharoubi, A. & Pien, H. 2000 Modeling dynamic behaviour of leaf temperature at three-dimensional positions to step variation in air temperature and light Trans. ASAE 43 1755 1766
Bot, G.P.A. & Van de Braak, N.J. 1995 Radiation 132 134 Bakker J.C. , Bot G.P.A. , Challa H. & Van de Braak N.J. Greenhouse climate control, an integrated approach Wageningen Pers Wageningen, The Netherlands
Buwalda, F. , Rijsdijk, A.A. , Vogelezang, J.V.M. , Hattendorf, A. & Batta, L.G.G. 1999 An energy efficient heating strategy for cut rose production based on crop tolerance to temperature fluctuations Acta Hort. 507 117 125
De Zwart, H.F. 1996 Analyzing energy-saving options in greenhouse cultivation using a simulation model Wageningen Agr. Univ Wageningen, The Netherlands PhD Diss.
Farquhar, G.D. & Von Caemmerer, S. 1982 Modelling of photosynthetic response to environmental conditions 549 587 Lange O.L. , Nobel P.S. , Osmond C.B. & Ziegler H. Water relations and carbon assimilation. Encyclopedia of plant physiology. New-series Physiol. Plant Ecol. II. Springer-Verlag Berlin, Germany
Farquhar, G.D. , Von Caemmerer, S. & Berry, J.A. 1980 A biochemical model of photosynthetic CO2 assimilation in leaves of C3 species Planta 149 78 90
Gal, S. , Angel, A. & Seginer, I. 1984 Optimal control of greenhouse climate: Methodology Eur. J. Oper. Res. 17 45 56
Gijzen, H. 1994 Ontwikkeling van een simulatiemodel voor transpiratie en wateropname en van een integral gewasmodel (in Dutch; Development of a simulation model for transpiration and water uptake and an integral crop model) DLO Rapport 18, AB-DLO Wageningen, The Netherlands
Goudriaan, J. & Van Laar, H.H. 1994 Modelling potential crop growth processes Kluwer Dordrecht, The Netherlands
Huber, L. & Gillespie, T.J. 1992 Modeling leaf wetness in relation to plant disease epidemiology Annu. Rev. Plant Phytopathol. 30 553 577
Jakobsen, L. , Brogaard, M. , Körner, O. , Enkegaard, A. & Aaslyng, J.M. 2005 The influence of a dynamic climate on pests, diseases and beneficial organisms: Recent research 127 134 Enkegaard A. Integrated control in protected crops, temperate climate IOBC wprs Bulletin Turku, Finland
Jones, H.G. 1992 Plants and microclimate Cambridge University Press Cambridge, U.K
Kempkes, F.L.K. & Van de Braak, N.J. 2000 Heating system position and vertical microclimate distribution in chrysanthemum greenhouse Agr. For. Meteorol. 104 133 142
Kim, S.H. & Lieth, J.H. 2003 Parameterization and testing of a coupled model of photosynthesis conductance for greenhouse rose crop Acta Hort. 593 113 120
Körner, O. 2004 Evaluation of crop photosynthesis models for dynamic climate control Acta Hort. 654 295 302
Körner, O. , Andreassen, A.U. & Aaslyng, J.M. 2006 Dynamic control of artificial lighting Acta Hort. 711 151 156
Körner, O. , Bakker, M.J. & Heuvelink, E. 2004 Daily temperature integration: A simulation study to quantify energy consumption Biosystems Eng. 87 333 343
Körner, O. & Challa, H. 2003a Design for an improved temperature integration concept in greenhouse cultivation Computers Electronics Agr. 39 39 59
Körner, O. & Challa, H. 2003b Process-based humidity control regime for greenhouse crops Computers Electronics Agr. 39 173 192
Körner, O. & Holst, N. 2005 Model-based humidity control of grey-mould in greenhouse cultivation Acta Hort. 691 141 148
Lund, H. 1995. The design reference year user manual, a report of Task 9: Solar radiation and pyranometer studies. Solar Materials Research and Development, International Energy Agency Solar Heating and Cooling Programme, Report No. IEA-SHCP-9E-1, Report No. 274, Thermal Insulation Laboratory, Technical University of Denmark, Denmark.
Lund, J.B. , Andreassen, A.U. , Ottosen, C.O. & Aaslyng, J.M. 2006 Effect of a dynamic climate on energy consumption and production of Hibiscus rosa sinensis L. in greenhouses HortScience 41 384 388
Ross, J. 1975 Radiative transfer in plant communities 13 56 Monteith J.L. Vegetation and the atmosphere Academic Press London, New York, San Francisco
Seginer, I. , Gary, C. & Tchamitchian, M. 1994 Optimal temperature regimes for a greenhouse crop with a carbohydrate pool: A modelling study Scientia Hort. 60 55 80
Stanghellini, C. 1987 Transpiration of greenhouse crops Wageningen Agr. Univ Wageningen, The Netherlands PhD Diss.
Tantau, H.J. 1998 Energy saving potential of greenhouse climate control Math. Comput. Simul. 48 93 101
Thornley, J.H.M. 1976 Mathematical models in plant physiology Academic Press London, U.K
Van Kraalingen, D.W.G. & Rappoldt, C. 1989 Subprograms in simulation models CABO Report 18, CABO Wageningen, The Netherlands
Van Pee, M. , Janssen, K. , Berckmans, D. & Lemeur, R. 1998 Dynamic measurement and modelling of climate gradients around a plant for micro environmental control Acta Hort. 456 399 406
Wang, S. & Deltour, J. 1999 An experimental model for leaf temperature of greenhouse-grown tomato Acta Hort. 491 101 106
Yang, X. 1995 Greenhouse micrometeorology and estimation of heat and water vapour fluxes J. Agr. Eng. Res. 61 227 238
Yu, Q. , Goudriaan, J. & Wang, T.D. 2001 Modelling diurnal courses of photosynthesis and transpiration of leaves on the basis of stomatal and non-stomatal responses, including photoinhibition Photosynthetica 39 43 51
Zhang, Y. , Jewett, T.J. & Shipp, J.L. 2002 A dynamic model to estimate in-canopy and leaf-surface microclimate of greenhouse cucumber crops Trans. ASAE 45 179 192
Zhang, Y. , Mahrer, Y. & Margolin, M. 1997 Predicting the microclimate inside a greenhouse: An application of a one-dimensional numerical model in an unheated greenhouse Agr. For. Meteorol. 86 291 297
