Effect of near-infrared-radiation reflective screen materials on ventilation requirement, crop transpiration and water use efficiency of a greenhouse rose crop

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Effect of near-infrared-radiation reflective screen materials on ventilation requirement, crop transpiration and water use efficiency of a greenhouse rose crop Cecilia Stanghellini a,*,1, Jianfeng Dai b,1, Frank Kempkes a,1 a b

Wageningen UR Greenhouse Horticulture, PO Box 644, 6700 AP Wageningen, The Netherlands College of Agriculture, Nanjing Agricultural University, Weigang No. 1, 210095 Nanjing, China

article info

The effect of Near Infrared (NIR)-reflective screen material on ventilation requirement,

Article history:

crop transpiration and water use efficiency of a greenhouse rose crop was investigated in

Received 18 June 2010

an experiment whereby identical climate was ensured in greenhouse compartments

Received in revised form

installed with either NIR-reflective or conventional material as internal movable screens.

2 August 2011

The NIR-filter reduced the energy load of the greenhouse by 8%. The high reflectivity of

Accepted 4 August 2011

the canopy in the NIR range caused the energy input of the greenhouse to be reduced by

Published online 13 September 2011

less than the properties of the material would suggest (25%). Both the ventilation requirement of the greenhouse and crop transpiration were reduced by the NIR-selective screen, consistently with the reduction in energy load of the greenhouse and the crop. The potential for commercial application of such material - either as movable screen or semi-permanent screen in addition to cooling - seems limited to high-tech greenhouse production in arid regions, where the reduction in cooling requirement could significantly lower water use for evaporative cooling. The potential for application of a photo-selective paint as alternative to the seasonal whitewash in low-tech greenhouses is obviously much larger. For it to make a difference, however, the NIR-reflectance must be much higher than the value of the present commercially available paints. ª 2011 IAgrE. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

High summer temperatures greatly limit greenhouse productivity. The common damage-reduction approach is to limit the amount of solar radiation that enters the greenhouse, either by whitewash or shadow screens, which have a high reflectivity (around 50%) across the whole solar spectrum. Such a reduction of Photosynthetically Active Radiation

(PAR) lowers photosynthesis (and thus crop growth). However, only about half of the energy of solar radiation is in the PAR range of 400e700 nm, with Near Infrared Radiation (NIR, 700e2500 nm) accounting for nearly all the rest. Filtering out NIR radiation would therefore reduce the energy load of the greenhouse by some 50%, without affecting crop photosynthesis and would be of great benefit to greenhouse growers, particularly those in warm climates. In addition, since

* Corresponding author. Tel.: þ31 317 483391; fax: þ31 317 418094. E-mail addresses: [email protected] (C. Stanghellini), [email protected] (J. Dai), [email protected] (F. Kempkes). 1 All authors contributed equally to the work. 1537-5110/$ e see front matter ª 2011 IAgrE. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.biosystemseng.2011.08.002

262

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

transpiration - being driven by the energy intake of the canopy - would be reduced, there is a great potential for increasing water use efficiency of crops. The potential for NIR-filtering greenhouse covers has been known for many years. The first detailed investigations of the cooling effect of a water film with NIR-filtering additives were carried out by Morris, Trickett, Vanstone, and Wells (1958). Following the first energy crisis, the potential of collecting the NIR-fraction of solar energy for night-time greenhouse heating was investigated by Van Bavel, Damagnez, and Sadler (1981), Chiapale, Van Bavel, and Sadler (1983) and Sadler and Van Bavel (1984). The principle was to pump a NIR-absorbing liquid through the cells of a hollow polycarbonate cover whenever cooling was needed, with the warm fluid being stored and eventually retrieved for heating. After these desk studies, experiments were done with a small, non-cropped greenhouse (Gale, Feuermann, Kopel, & Levi, 1996), but no larger scale results have been reported, as far as we know. Implementation faced technical problems, the development of algae in the roof cells being an acknowledged one. More recently, the research has focused on developing solid NIR-filtering materials, such as plastic films or coated glass for greenhouses covers (Abdel-Ghany, Kozai, & Chun, 2001; Verlodt & Verschaeren, 1997) or sheets to be used as movable screens (Runkle, Heins, Jaster, & Thill, 2002a; Tanaka, 1997). NIR-filtering paints for use as whitewash have been developed as well (Von Elsner & Xie, 2003), although the spectral selectivity of the materials commercially available is still far from ideal. A filtering material can work through absorption or reflection. Obviously, absorption would warm up the material and a large fraction of the absorbed energy would end up in the internal greenhouse environment anyhow, as convection and thermal radiation. Reflection would be much more efficient but is more difficult to achieve, usually through a stack of layers of different optical properties to ensure interference in the desired range (Hemming, Kempkes, van der Braak, Dueck, & Marissen, 2006a; Hoffmann & Waaijenberg, 2002; Sonneveld, Swinkels, Bot, & Flamand, 2010). The overall effect of several types of application of an NIRfilter (permanent, seasonal, movable within and without a greenhouse) on Dutch greenhouse production was investigated within our group by Hemming, van der Braak, Dueck, Elings, and Marissen (2006b) through a greenhouse climate model. The main conclusions were: (1) a permanent NIRfiltering cover reduces the ventilation requirement in summer, thereby allowing more effective carbon dioxide (CO2) fertilization, and hence increased production. However, the year-round energy consumption would increase by 10%, because of the reduced input of solar energy, particularly in winter; (2) a seasonal/movable filter would be beneficial only when the ventilation of the greenhouse would not be hampered; (3) the balance of costs and benefits depends on the selectivity of the filter. This analysis was extended by Kempkes, Stanghellini, and Hemming (2008a) to passive, unheated greenhouses in mild winter climates, who concluded that the cooling caused by a permanent filter in winter would more than offset the gain of the summer cooling. In addition, the reduced ventilation requirement may result in lower CO2 concentrations - in the absence of CO2

supply, which is still a rarity in unheated greenhouses. In short, there is potential for such materials installed as movable screens in high-tech greenhouses, in such a way as not to hamper ventilation, and as a seasonal filter (either as paint or outside screen) on greenhouses in mild-winter regions, keeping in mind that absence of CO2 fertilisation may off-set some of the benefit. In spite of the novelty of such materials, some experimental works have been published on their effects (e.g. Runkle et al., 2002a, 2002b; Arcidiacono, D’Emilio, Mazzarella, & Leonardi, 2006; Garcia-Alonso et al., 2006; Impron, Hemming, & Bot, 2008) and on the results from application of photo-selective paints by commercial firms in The Netherlands (Van Rijssel, 2006; Van Staalduinen, 2008; Vegter, 2007). In all these works, however, the effect of the screen on the crop is mediated through the effect of the screen on the internal climate. The evaluation of such a filter against other cooling methods must account for any possible direct effect of the screen properties on crop productivity. This holds true also for growers operating the low-tech greenhouse facilities typical of warmer conditions. As they have little control on the inner climate resulting from whitewash, evaporative cooling, or such a photoselective filter, there is a need to separate (and be able to predict) the effect of the cooling action on the inner climate from any direct effect on the crop. The purpose of this study was to fill this gap, through an experimental design that ensured the same climate (temperature, CO2 concentration and PAR radiation) in compartments installed either with a NIR-reflective or a conventional screen. We selected roses for the test, since it is a crop for which even in The Netherlands there is a widespread use of shadow screens, whitewash and cooling methods such as high-pressure fogging or roof spraying. In addition, it is a high-value crop, a prerequisite for cost-effectiveness of a rather expensive fitting such as this one.

2.

Materials and methods

2.1.

Greenhouse and experimental set up

The experiment was carried out in 4 identical compartments (144 m2 each) of the 1 ha greenhouse research facility of Wageningen UR Greenhouse Horticulture, Bleiswijk (52 N, 4.5 E), The Netherlands. The greenhouse has four rows of 13 compartments each, separated by three North-South corridors, and is surrounded on all sides by a corridor (Fig. 1). Each Venlo-type compartment (15 m long) is composed of 2 EeW oriented, 4.8 m wide, spans. The height of the gutter is 5.5 m and the roof angle is 22 . Each span is equipped, on both sides of the ridge, with continuous roof vents, 14 m long and 1.28 m wide, fitted with insect-proof nets. The present experiment was performed in four compartments aligned EeW, which were, therefore, separated by the corridors and were not “shading” each other. The neighbouring compartments (to the South and North of each one) and the corridors were fitted with standard roofs. Internal movable screens were installed into the four compartments, two with NIR-reflective material (hereafter NIR, 3M, USA) and two (for reference) with a screen that was the one with the most comparable properties in the

263

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

Fig. 1 e Schematic representation of the greenhouse and the place of the experimental compartments in it. The thin lines indicate the position of each gutter (separation between two spans). Each compartment was two spans (9.60 m) wide and 15 m long. The plot on the right gives the distribution of wind direction during the experimental period.

1 0.8 0.6

NIR

0.4

reflection absorption

02

transmission 0 1

0.8 0.6

REF 0.4 0.2

reflection absorption transmission

0 300

500

700

900

1100

1300

1500

1700

1900

2100

2300

2500

nm

Fig. 2 e Spectral properties for perpendicular light of the two materials installed as movable screens.

264

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

Fig. 3 e Scheme of the screen installations. For each greenhouse span there were four rolls, each sliding parallel to the roof slope, two in front of the continuous zenithal vents (A and B) and two in front of the lower slopes (C). The vents were fitted with insect nets. The A, B and C screens were controlled independently, the ones in front of the vents being folded in the measure that the flap was open. The screen was fully folded when the flap reached the maximal opening angle q of 52 .

PAR range, among those commercially available (hereafter REF, ILS ultra, Ludvig Svensson). Interference with the surrounding compartments was limited as much as possible by the use of white, insulating screens on the sidewalls. The spectral transmittance and reflectance for perpendicular light of the photo-selective and the reference material are

shown in Fig. 2. In contrast to the photo-selective foil, which is clear, the reference screen is woven and has some light diffusivity. A better indicator of performance was measurements through an Ulbricht integrating sphere (possible only up to 1100 nm). The overall transmittance in the PAR range of the two materials was 87% (NIR) and 84% (REF). As Hemming, Dueck, Janse, van Noort, and van Henten (2008) have shown that, in terms of crop production, diffusivity may make up for some light loss, we expected no consequence from the small difference in PAR transmissivity. As the figure shows, the transmittance of the 3M material has a sharp cut-off in the wavelength range between 840 and 1120 nm for perpendicular light, whereas at larger incidence angles there is an increasing reduction in the red (above 650 nm - data not shown). Altogether, accounting for the spectral energy content of sunlight, this material transmits about 50% of the NIR radiation and 75% of global radiation. To avoid interference between the screen and ventilation, 4 rolls of the material were installed within each span, sliding parallel to the roof slope, two along the non-opening part of the North and South slopes, and two along the vents, Fig. 3. The purpose of the climate control was to maintain the same temperature in all cases, by delaying ventilation and taking as much advantage as possible of the NIR-filter, even when ventilation was required. Therefore, operation of the screen was controlled in 3 independent segments: A, in front of the North-facing vent; B, in front of the South-facing vent and C, both rolls facing the non-moving part the roof. The operation of the segments was controlled by the climate computer as

Fig. 4 e Operation of the screens during a sunny day (sun radiation, dotted line, on the right hand axis). Each grey area indicates closure of one group of screens (A, B and C, refer to Fig. 3), the line including the shaded area indicating the resulting screen deployment. When the greenhouse air temperature (blue, on the left hand axis) reached the ventilation setpoint (21  C, lowest horizontal line), all screens were fully deployed (A, B and C). In spite of the clear effect, air temperature remained above the ventilation line, so that the North-facing vents had then to open gradually and the corresponding screens (A) were folded by the same amount, until (around 11:00) full vent opening was not enough to control air temperature below 27  C (higher horizontal line), and the South-facing vents were gradually open and screens B gradually folded. The reverse happened in the afternoon, when screen A was gradually deployed again when the greenhouse cooled below 27  C. All screens were folded when sun radiation dropped to 50 W m2, even if the temperature was still above 21  C. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

265

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

follows: all screens were folded whenever solar radiation was below 50 W m2, since they could not have a cooling effect. Above that level, as soon as greenhouse air temperature reached within 0.3  C of the ventilation set-point, all screens would be deployed. When, in spite of this, the greenhouse air temperature proceeded to exceed the ventilation set-point (21  C), the North-vents were gradually opened (proportional control with a P-band of 6  C) and the screens facing them (A) folded by the same amount. The South-vents and the screens facing them would be operated, in the same fashion, only when the North-vents were fully open and the temperature would exceed 27  C, Fig. 4. The reverse procedure was activated when the greenhouse cooled. The purpose was to maximize effectiveness of the screens by keeping the Southfacing ones deployed as long as possible, so as to exclude most (direct) radiation.

2.2.

Crop and crop management

Rooted cuttings of cut roses (Rosa hybrida, cv. Passion) were planted on March 11th 2008 in rockwool slabs placed on EeW oriented suspended gutters, with a plant density of 5.77 plants m2. The plants were grown following the ‘bending’ technique, which consists of bending over the primary stem and all stems that were not considered useful to flower production. This technique allows more leaf area for sustaining photosynthesis and increases the contribution of the canopy transpiration to greenhouse cooling (e.g. Gonzalez-Real & Baille, 2000). By the time of the first harvest (June 9th) the crop was about 1.5 m high and the modal length of harvested stems exceeded 70 cm. The plants were irrigated by means of a drip system, which was automatically controlled by a fertigation computer. Water and nutrient supply was based on outside solar radiation and crop age, and was identical for all compartments. The pH of the nutrient solution was adjusted to 5.2 and the EC value was maintained around 1.6 dS m1. The climate management (identical in all compartments) was as follows. The set point for heating was 19  C at night and 20  C at daytime and the set point for ventilation was 21  C. The relative humidity inside was controlled by a high-pressure fogging system, the set point decreasing from 85% immediately after transplanting, to 75% up to July 5th, when the fogging was discontinued. Daytime inside CO2 concentration was controlled to 1000 ppm, although the capacity of the supply system sometimes proved insufficient under high ventilation rates. Artificial light (Philips SON-T lamps, 100 mmol m2 s1) was switched on whenever outside radiation was lower than 100 W m2 during the 17 h preceding sunset. Once on, the lamps were switched off if the radiation exceeded 150 W m2. In order to ensure initially similar crop development, the screening treatments were not applied until April 29th, 2008 and the experiment was terminated on October 31st, 2008.

2.3.

Measurements

The overall transmittance of the greenhouse for solar radiation, both in the PAR and global range, was determined as the average of several spatial samples taken at flower level on July 8th and September 24th 2008, in both cases under diffuse light

conditions, once with all screens deployed and once with all screen folded. One set of instruments, a quantum sensor and a pyranometer (sensing area about 1 cm2 each), was used to measure inside the greenhouse, and another set was used outside at the same time. The NIR radiation flux was determined as the difference between global and PAR radiation of each sample, and the transmittance determined accordingly. Concurrently, canopy reflectance was determined through a similar set of sensors, facing downward. Throughout the experiment, the standard climate measurements (air temperature, relative humidity, global radiation, wind speed and wind direction) were recorded by the weather station of the research facility, located at a height of about 8 m. In addition, the following were recorded for each compartment: dry and wet bulb temperatures and CO2 concentration, through sensors in a ventilated box 1.3 m above ground; PAR intensity through a quantum sensor 10 cm above the canopy; and canopy temperature through an infrared thermometer 20 cm above the canopy, facing slightly downwards, so that the viewing area was about 1.5 m2. The following actuators of the climate control were also recorded: the calculated opening (%) of all screens and ventilators, the on/off status of the artificial light and of the CO2 supply, and the temperature of the heating pipes. Crop transpiration was determined through six load cells (capacity of 250 kg, resolution of 0.1 g), installed in each compartment, supporting two coupled gutters, each a halfrow long, for a total of 60 plants. The drain from the weighed gutters was recorded through a tipping spoon counter, starting on July 5th, 2008. As the rockwool slabs were packaged in plastic, any weight decrease of the gutters had to come from crop transpiration, once harvest and drain were considered. Accordingly, crop transpiration was determined as the derivative of the weight of the gutter, after adding the drain to it. Sudden weight increases (irrigation events) and decreases (harvests) were filtered out. For the period before the drain measurement was available, we discarded the values following an irrigation event. All data (weather, greenhouses and gutters) were recorded each minute, and the 5-min means were saved. In view of the signal to noise ratio, we estimated that transpiration data were reliable on time intervals of 30 min at least.

Table 1 e The overall transmittance properties of the greenhouse compartments and crop reflection for diffuse light. Properties in the PAR and global range were measured, the NIR flux density was calculated as the difference of the two, and the properties in the NIR range were estimated accordingly. Fully folded Fully deployed Crop reflection NIR PAR (%) NIR (%) Global (%) PAR/global REF PAR (%) NIR (%) Global (%) PAR/global

44 46 45 0.51 44 45 46 0.51

38 29 34 0.59 36 38 37 0.51

5 45 24 5 45 24

266

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

The number and fresh weight of harvested stems were recorded for each compartment, each harvest day. In addition, 3 plants per compartment (except roots) were destructively sampled at 4-week intervals to determine the leaf area, with a leaf area metre LI3000, and plant dry weight, after drying for 48 h at 80  C.

3.

Results

3.1.

Light transmittance and heat load reduction

The overall diffuse transmittance in the global and PAR ranges, measured at flower level in the compartment, with both folded and deployed screen is shown in Table 1. The relatively small effect of the NIR-selective screen on transmitted global radiation (and thus the heat load of the greenhouse) may seem surprising, in view of the properties of the material, described above. This is due to the much higher reflectance of leaves for NIR rather than PAR radiation (Table 1), which is known (e.g. Stanhill, Fuchs, Bakker, & Moreshet, 1973; Lillesand & Kiefer, 2003) but whose importance is often overlooked. This obviously reduces the effectiveness of the screen, since part of the NIR radiation reflected by the canopy is reflected again at the internal screen surface, and so on.

3.2.

Greenhouse climate

The experiment was designed to have the same climate inside all compartments, in spite of the treatment, something that we largely achieved. What we did not consider in the experimental design was that the 4 experimental compartments were in a row (from West to East: two NIR-screened and two reference) along the prevailing wind direction (WeE, Fig. 1) spanning the whole dimension of the greenhouse. De Jong (1990) has observed that the ventilation characteristics of the most windward compartments of a large multi-span Venlo-type greenhouse are different from the central ones, due to the larger static pressures. Indeed we observed that smaller window openings were usually required to ensure the same temperature, even under the same type of screen (Table

Table 2 e Daytime temperature ( C) and vents opening (% of the maximal opening angle of 52 ) in the four compartments. The second line in each “vents” group is the mean of the two replicates of each treatment. The vents were controlled in order to warrant the same temperature in all compartments. The different ventilation characteristics of the outer compartments resulted in a smaller opening angle being needed to ensure the same temperature, even for the same treatment. Different letters indicate significant differences between treatments for P < 0.05. NIR outer NIR inner REF inner REF outer LSD Temperature North-vents South-vents

24.36a 37.01a 40.50a 1.95b 1.90a

24.37a 43.99c 1.86a,b

24.59c 46.32d 44.88a 2.34c 1.98a

24.53b 43.45b 1.73a

0.044 0.53 16.22 0.17 1.10

Table 3 e Average daytime values of the most relevant climate factors over the experimental period (April 28th to Oct 31st, 2008), unless otherwise indicated. Different letters indicate significant differences between treatments for P < 0.05.

Air temperature Canopy temperature1 PAR2 CO2 concentration Absolute humidity3 Absolute humidity4 Vapour deficit3 Vapour deficit4 Operation time fogging3



C C mol m2 d1 ppm g kg1 g kg1 g kg1 g kg1 min d1 

NIR

REF

LSD

24.37a 24.50b 13.3b 901b 14.69b 13.85a 5.28a 5.08a 102.3

24.59b 24.41a 12.9a 869a 14.60a 14.02b 5.75b 5.14a 92.3

0.044 0.050 0.19 3.2 0.042 0.040 0.048 0.069 3.38

1

from May, 22nd from June, 2nd 3 up to July, 5th, fogging was used. 4 from July, 5th, fogging was not used. 2

2). Since the screen opening was coupled to the window opening, this introduced a variation within our replicate treatments, see Table 2. Table 3 gives the daytime means of relevant climate variables in the two inner compartments, for the experimental period. There was no treatment at night and indeed there were no differences in climate. The sheer number of data (more than 30 thousand lines) and the precision of the climate control ensure that even differences below the accuracy of the measurements appear to be statistically significant. The small difference in air temperature between the two treatments follows from the control, as can be seen from Fig. 4: the temperature set-points for the two vents (and facing screens) were the same, but the energy load was not. Most days of that summer were not as sunny as the one shown, and the differences were usually much smaller. Vapour concentration was not controlled and was slightly lower in the NIR than in the REF compartment after the fogging was discontinued, which must follow from the lower transpiration rate, in spite of the lower ventilation rate (discussed below). The canopy temperature was the same for both, within the error of the sensor. Canopy temperature follows from the energy balance, and the lower transpiration in the NIR compartment apparently offsets the lower air temperature. The small difference in accumulated PAR radiation matches the screen properties shown in Table 1. Finally, the capacity of the CO2 supply installation was not great enough to maintain the desired concentration of 1000 ppm under high ventilation rates. The fact that there was a greater need for ventilation in the REF than in the NIR compartment explains the small difference, 32 ppm, that was observed between the two.

3.3.

Ventilation requirement

Indeed, the ventilation flow rate in the NIR compartment was lower than that in the REF compartment. One sunny day (July 25th, 2008) and one cloudy day (August 28th, 2008) were selected for calculating the ventilation flow, following the method of De Jong, 1990, which accounts for the geometry of

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

fact, as the vents had to open more in the sunny day, the screens were deployed less, so that there was less difference between the treatments. The reduction of 8.8% in the ventilation rate brought about by the NIR-selective screen is perfectly consistent with the difference in energy load (Table 1). Altogether, as Kempkes, Stanghellini, Hemming, and Dai (2008b) have pointed out, since natural ventilation is the single cooling method in most greenhouses, any application of NIR-filtering material in such a way that the ventilation would be prevented would be counterproductive, in view of its relatively small reduction of the ventilation requirement. Obviously, in the presence of other cooling means than natural ventilation, such as pad and fan cooling, its effect on energy load would decrease the cooling requirement.

0.06 Sunny day NIR (m3 m-2 s-1)

Cloudy day 0.04 y = 0.93 x 0.02

0

3.4.

0

0.02

0.04

267

Crop transpiration

0.06

REF (m3 m-2 s-1) Fig. 5 e The calculated specific ventilation flow rate in the two compartments for a sunny and a cloudy day (July 25th and August 28th, 2008, respectively). The line is the best-fit over all points.

the vents and requires as inputs: wind speed; vent opening angle and temperature difference between in- and outside, see Appendix. The method yields the ventilation rate of an “infinitely large” Venlo glasshouse, which definitely is not our case. However, it seems reasonable to assume that, as the two compartments can be considered of equal size and are both fairly central in the greenhouse, their ventilation rate will diverge equally from the “theoretical” one. Therefore we compare the resulting ventilation flows in Fig. 5. The slopes (93.4% and 91.2%, respectively for the sunny and cloudy day) are significantly different within a confidence of 0.05. It may surprise that the ventilation rate was reduced by the NIR screen more for the cloudy day than that for the sunny day. In

Fig. 6 shows the hourly averages of transpiration in all compartments for five days. There is no difference between treatments at night-time, which is logical since there was no treatment. The slope of the daily water uptake NIR vs REF is 94% (r2 0.98) for the period after July 5th.

3.5.

Crop growth and production

Given the experimental design and the success in ensuring the same climate, any difference in crop growth and development had to be caused by the screen properties, or by the difference in water uptake. We did not observe any difference in LAI, nor in any parameter of crop growth (Table 4). The crop water use efficiency (WUE) is often defined as crop dry weight gain per kg water used. For a rose crop, as the value of yield is mainly dependent upon the fresh weight (FW) of the harvested stems, water productivity is defined as the number of marketable stems or harvest fresh weight produced per kg water (e.g. Duchein, Baille, & Baille, 1995). The results in Table 4 show that there was no effect of NIR reflective screen treatments on biomass production, number and fresh weight of harvested stems, nor on their length (not shown), while

Fig. 6 e Trend of crop transpiration for 5 days in the two compartments (g mL2 hL1, left axis) and corresponding sun radiation (W mL2, right axis). Data shown are running averages over 1.5 h. Artificial light was supplied before sunrise and whenever sun radiation was below 100 (ON) or 150 (OFF) W mL2.

268

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

Table 4 e Number of stems, fresh and dry weight of harvest and LAI for the period from July 13th to October 13th, 2008, respectively the first and last harvest of the experiment. Transpiration is from July 5th to Oct 31st. Different letters indicate significant differences between treatments for P < 0.05. Harvest number (stems m2)

Harvest FW (kg m2)

Biomass produced (g DW m2)

LAI.

Transpiration (mm d1)

51.3  4.73a 54.6  4.93a

2.098  0.189a 2.104  0.184a

699.98a 700.14a

3.85  0.28a 4.36  0.44a

1.355  0.029a 1.429  0.035b

there was a significant reduction in crop transpiration under the NIR screen. Consequently water productivity was increased as much as the water uptake was decreased.

4.

Discussion

4.1.

Combined reflectance of cover and crop

An estimate of the overall effect of the relatively high reflectivity of the crop for NIR can be made as shown by Fig. 7. The combined reflectance r of two superimposed layers with reflectances of r1 and r2, respectively, is given by: r ¼ r1 þ r2 ð1  r1 Þ2

N X n¼0

rn1 rn2 ¼

r1 þ r2  2r1 r2 1  r1 r2

In our case, the canopy had a reflectance of 24% for global radiation (Table 1) and the photoselective screen of 25%, whereas the reflectance of the reference was 15% (Fig. 2). Calculating the equation above for the two cases, shows that if the transmittance in the REF compartment (with the screen deployed) was 37% (Table 1), it should have been 33.5% in the NIR, which is close to the 34% we measured. More generally, Fig. 8 shows the net effect of a reflecting cover superimposed on a reflecting crop. The effective increase in reflectance brought about by the cover is largest when the reflectivity of the crop is small, that is with small plants and incomplete soil cover. For instance, the standard growing cycle in greenhouses in the Mediterranean region is to plant in August and apply heavy whitewash (cutting out some 50% of solar radiation). A filter like this one would

probably allow for a much lighter whitewash and thus a higher fraction of PAR would reach the crop. On the other hand, a law of “increasing returns” holds for increasing the reflectance of the cover over a reflecting crop. For instance, with a soil-covering crop reflecting 50%, to increase the reflectance of the cover from 30 to 40% would increase the combined reflectance by only 3.8%, whereas the same increase from 70 to 80% would have an effect of 6.4%. This means that there is little benefit to be expected from marginal improvements of poorly reflecting properties. Indeed, Impron et al. (2008), after comparing the effect, in tropical conditions, of greenhouse covers with a NIRreflectance increasing from 19 to 36%, observed a far larger effect of ventilation and leaf area than of the cover properties on temperature in the greenhouses. Their conclusion was that “the variation in cover properties of the experimental greenhouses was too small to show effects”, which has been demonstrated here. Obviously, as the reflectance of the material used here is relatively high, there would be some benefit in improving upon it. An additional observation here is that the perceived benefit of NIR-reflection upon NIR-absorption may be easily overestimated.

1

0.8 additional reflectance

NIR REF

0.6

0.4

0.2

0 0 Fig. 7 e Schematic representation of the behaviour of radiation between two parallel planes of partial reflectance r1 and r2, respectively. The drawing is for light of one particular angle of incidence, but it applies as well to any distribution of incidence angles, provided the reflectance coefficients apply to that distribution.

0.2

0.4

0.6

0.8

1

reflectance cover Fig. 8 e Net effect of a cover on a crop having reflectance indicated by the lines as follows: full line 10%; long dashes 30% and short dashes 50%. On the x-axis is the reflectance of the cover, and on the y-axis the gain with respect to the reflectance of the crop.

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

4.2.

Crop transpiration

We have proven that crop transpiration is reduced by about the same amount as the radiation on the crop is reduced. The fact that it is reduced by slightly less (6% against 8%) may be easily attributed to inaccuracy of the measurements, since LAI is, if anything, smaller in the NIR compartment, although the difference is not significant. Nevertheless, as may be deduced from the PenmaneMonteith equation, crop transpiration is the sum of an “equilibrium” component which is proportional to canopy net radiation and an “imposed” component which is proportional to air vapour saturation deficit, so that a reduction in radiation translates into an equal reduction of transpiration only in a perfectly closed environment (McNaughton & Jarvis, 1983). The fact that the reduction in radiation was not fully matched by the reduction in transpiration attests to the ventilation of our glasshouses. In an experiment in the Mediterranean region, Baille, Kittas, and Katsoulas (2001) have shown that the transpiration of a rose crop was slightly higher under whitewash than without. What they found was that, whereas without whitewash the canopy was warmer than the air, it was colder than air under whitewash, so that the sensible heat brought to the canopy more than made up for the reduction in radiation, and more energy was available for transpiration. Partial stomatal closure, caused by water stress in non-whitewash conditions, may have been the cause. In spite of our efforts not to have an effect on climate (and to the degree that such small differences we have observed are indeed real), the NIR-selective screen cooled the air more than it cooled the canopy (0.2 against 0.1  C), so that under the NIR-filter the crop was slightly warmer than the air, whereas it was slightly colder in the reference. Monteith (1981) pointed this out for evaporative cooling in greenhouses. This follows from “hydraulic feed-back” in the semi-closed environment of a greenhouse (Aubinet, Deltour, De Halleux, & Nijskens, 1989; Stanghellini & De Jong, 1995). That is, the crop transpires less, the air gets dryer, so the crop transpires more: or, in the case of evaporative cooling, the air gets more humid, the crop transpires less, the crop gets warmer, the crop transpires more. The presence of such feedbacks ensure that any change in the environment translates into a less than proportional effect on crop transpiration, all other factors being the same, of course. The operation of the fogging (with a set-point based on relative humidity), obviously would mitigate (or interrupt) the hydraulic feed-back, and we have indications that the difference in transpiration between the treatments was indeed much larger when the system operated than afterwards. However, in this period the soil cover was still incomplete, which would increase the difference between the treatments (Fig. 7) and the absence of a direct measurement of drain made the measurement of the transpiration less accurate. Therefore we prefer not to give undue relevance to this observation. Altogether, the reduction in potential transpiration that we achieved without affecting any of the variables that also affect photosynthesis was bound to result in higher water use efficiency. An easier way to achieve the same is through reduced vapour pressure deficit (Li, Marcelis, & Stanghellini, 2004), which also does not affect photosynthesis insofar as water stress is

269

avoided. Indeed, it is known that water use efficiency of crops increases when vapour deficit decreases. The other way round, raised carbon dioxide levels increase water use efficiency by increasing photosynthesis without increasing transpiration.

5.

Conclusion

A NIR-reflective material and a reference with similar PAR transmissivity were used as internal movable screens in 4 greenhouse compartments, in order to investigate the effect of the photo-selective material on greenhouse ventilation requirement and production, transpiration and water use efficiency of a rose crop. The experimental design (identical climate in spite of the treatment) prevented any effect mediated through greenhouse climate (such as temperature, for instance). The main conclusions that may be drawn are:  The NIR-filter reduced the energy load of the greenhouse by 8%. The high reflectivity of the canopy in the NIR range causes the energy input to the greenhouse to be reduced by less than the properties of the material would suggest (25%).  Both the ventilation requirement of the greenhouse and crop transpiration were reduced by the NIR-selective screen. The amount of reduction was consistent with the reduction in energy load of the greenhouse and the crop.  The rather unusual experimental conditions whereby net radiation to the canopy was reduced without concurrent reduction in photosynthetically active radiation, demonstrated that transpiration of a well-watered crop may be reduced without necessarily reducing production. Economic water use efficiency of the crop was increased correspondingly to the reduction of transpiration. In short, the effect of a NIR-filter in the greenhouse cover should not be overstated. The potential for commercial application of such materialeeither as movable screen or semi-permanent screen in addition to coolingeseems limited to high-tech greenhouse production in arid regions, where the reduction in cooling requirement could significantly lower water use for evaporative cooling. The potential for application of a photo-selective paint as an alternative to the seasonal whitewash in low-tech greenhouses is obviously much larger. As the traditional cropping cycle in greenhouses in the Mediterranean region starts with small plants in August, the incomplete soil cover by the crop would make the NIR-filter most effective when it is most needed. For it to make a difference, however, the NIRreflectance must be much higher than the value of the paints that are commercially available at present.

Acknowledgements This research was funded by the Dutch Ministry of Agriculture, Nature and Food Quality (LNV); the Dutch Product Board for Horticulture (project 13287) and the European Commission (FP7: EUPHOROS, grant 211457). The photoselective material

270

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

was kindly provided and shipped by 3M, St. Paul, MN, USA, no conditions attached. This work was done while Dai Jianfeng was on sabbatical in Holland, his stay financed by The Netherlands Organisation for International Cooperation in Higher Education (NUFFIC, under a Sino-Dutch bilateral agreement) and by the Graduate School on Production Ecology and Resource Conservation of Wageningen University. We are grateful to Dott. Mirco Romani (University of Pisa, Italy) for his help with measurements during the experiment, and to Dr Silke Hemming for advice before, during and after the experiment. Our colleague Drs Vida Mohammadkhani measured repeatedly the optical properties of the materials. Thanks are also due to the greenhouse staff of Wageningen UR Greenhouse Horticulture: Gerard van den Broek, Fred van Leeuwen, Peter Schrama and Piet Koornneef, for crop management and technical support.

Appendix. The calculation of the ventilation rate De Jong (1990) parametrised a distributed-pressure ventilation model particularly for large multi-span greenhouses of the Venlo type such as ours. Hereafter we give his equations such as we have used for calculating the ventilation rate, Fig. 5. The total air exchange rate 4vent is the sum of two components: ventilation through the openings and leakage ventilation: fvent ¼ fopenings þ fleak

m3 m2 s1

(1)

m3 m2 s1

(2)

with: fleak ¼ 0:00013,maxðu; 0:25Þ 1

where u is the wind speed, m s . Superposition of the wind and buoyancy pressure vector components leads to an expression for the ventilation through the openings: 0:5  fopenings ¼ 0:5 N f2wind þ f2temp

m3 m2 s1

(3)

where N is the number of openings per m2 greenhouse. The wind-driven ventilation flow through a roof window is the sum of the flow through the triangular sides of the opening and its front, to a ventilation function G:  each proportional  (4) fwind ¼ Fwind Glength þ Gwidth uAwindow m3 s1 where Awindow is the area of the window, m2. For our window configuration and with an opening angle q, degrees, the ventilation functions can be written as:    q Glength ¼0:0229 1  exp  and 21:1   q Gwidth ¼ 0:0019qexp  211:1

e

(5)

In order to account for the additional pressure on the sidewalls of the greenhouse, De Jong introduced the factor Fwind that in fact increases the total greenhouse area, Agreenhouse, m2, with an effective “belt”, Abelt, along the perimeter of the greenhouse: Fwind ¼ and

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Agreenhouse þ Abelt Agreenhouse

e

(6)

    Abelt ¼ 0:0607p  hgutter p þ 4 0:0607p  hgutter

m2

(7)

where p and hgutter are the perimeter of the greenhouse and the height of the side wall, respectively. The ventilation through a roof window, caused by buoyancy is calculated through: Lwindow 1:5 Hwindow ½gbDT0:5 ftemp ¼ 0:6 3

m3 s1

(8)

where 0.6 is a coefficient to account for the discharge of energy caused by friction in the opening, Lwindow is the length and Hwindow the height of the front opening, which is a function of q; g and b are the gravitation constant and the thermal expansion coefficient of air, respectively, and DT indicates the difference in temperature between the greenhouse, which is warmer, and outside. Finally, substitution of Eq. (4) and Eq. (8) in Eq. (3), and of the resulting equation and Eq. (2) into Eq. (1) gives the equation used for the ventilation flow.

references

Abdel-Ghany, A. M., Kozai, T., & Chun, C. (2001). Evaluation of selected greenhouse covers for use in regions with a hot climate. Japanese Journal of Tropical Agriculture, 45, 242e250. Arcidiacono, C., D’Emilio, A., Mazzarella, R., & Leonardi, C. (2006). Covering materials to improve greenhouse microclimate during summer in hot climates. Acta Horticulturae, 719, 247e254. Aubinet, J., Deltour, D., & De Halleux Nijskens, J. (1989). Stomatal regulation in greenhouse crops: analysis and simulation. Agricultural and Forest Meteorology, 48, 21e44. Baille, A., Kittas, C., & Katsoulas, N. (2001). Influence of whitening on greenhouse microclimate and crop energy partitioning. Agricultural and Forest Meteorology, 107, 93e106. Chiapale, J. P., Van Bavel, C. H. M., & Sadler, E. J. (1983). Comparison of calculated and measured performance of a fluid-roof and a standard greenhouse. Energy in Agriculture, 2, 75e89. De Jong, T., 1990. Natural ventilation of large multispan greenhouses, PhD Thesis, Wageningen, the Netherlands: Agricultural University. Duchein, M.-C., Baille, M., & Baille, A. (1995). Water use efficiency and nutrient consumption of a greenhouse rose crop grown in rockwool. Acta Horticulturae, 408, 129e135. Gale, J., Feuermann, D., Kopel, R., & Levi, S. (1996). Liquid radiation filter greenhouses (LRFGs) and their use of low quality hot and cold water, for heating and cooling. Acta Horticulturae, 440, 93e98. Garcia-Alonso, Y., Espi, E., Salmeron, A., Fontecha, A., Gonzalez, A., & Lopez, J. (2006). New cool plastic films for greenhouse covering in tropical and subtropical areas. Acta Horticulturae, 719, 131e138. Gonzalez-Real, M. M., & Baille, A. (2000). Changes in leaf photosynthetic parameters with leaf position and nitrogen content within a rose plant canopy (Rosa hybrida). Plant, Cell and Environment, 23, 351e363. Hemming, S., Kempkes, F., van der Braak, N., Dueck, T., & Marissen, N. (2006a). Greenhouse cooling by NIR-reflection. Acta Horticulturae, 719, 97e105. Hemming, S., van der Braak, N., Dueck, T., Elings, A., & Marissen, N. (2006b). Filtering natural light at the greenhouse covering - better greenhouse climate and higher production by filtering out NIR. Acta Horticulturae, 711, 411e416.

b i o s y s t e m s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 6 1 e2 7 1

Hemming, S., Dueck, T. A., Janse, J., van Noort, F. R., & van Henten. (2008). The effect of diffuse light on crops. Acta Horticulturae, 801, 1293e1300. Hoffmann, S., & Waaijenberg, D. (2002). Tropical and subtropical greenhouses e a challenge for new plastic films. Acta Horticulturae, 578, 163e171. Impron, I., Hemming, S., & Bot, G. P. A. (2008). Effects of cover properties, ventilation rate, and crop leaf area on tropical greenhouse climate. Biosystems Engineering, 99(4), 553e564. Kempkes, F., Stanghellini, C., & Hemming, S. (2008a). Cover materials excluding near infrared radiation: what is the best strategy in mild climates? Acta Horticulturae, 807, 67e72. Kempkes, F., Stanghellini, C., Hemming, S., & Dai, J. (2008b). Cover materials excluding near infrared radiation: effect on greenhouse climate and plant processes. Acta Horticulturae, 797, 477e482. Li, Y. L., Marcelis, L. F. M., & Stanghellini, C. (2004). Plant water relations as affected by osmotic potential of the nutrient solution and potential transpiration in tomato (Lycopersicon esculentum L.). The Journal of Horticultural Science & Biotechnology, 79(2), 211e218. Lillesand, T. M., & Kiefer, R. W. (2003). Remote sensing and image interpretation (4th Ed).. USA: John Wiley & Sons, Inc. Publisher. McNaughton, K. G., & Jarvis, P. G. (1983). Predicting effects of vegetation changes on transpiration and evaporation. In T. T. Kozlowski (Ed.), Water deficit and plant growth, Vol. 7 (pp. 1e47). New York: Academic Press. Monteith, J. L. (1981). Evaporation and surface temperature. Quarterly Journal, Royal Meterological Society, 107(451), 1e27. Morris, L. G., Trickett, E. S., Vanstone, F. H., & Wells, D. A. (1958). The limitation of maximum temperature in a glasshouse by the use of a water film on the roof. Journal of Agricultural Engineering Research, 3, 121e130. Runkle, E. S., Heins, R. D., Jaster, P., & Thill, C. (2002a). Environmental conditions under an experimental near infrared reflecting greenhouse film. Acta Horticulturae, 578, 181e185.

271

Runkle, E. S., Heins, R. D., Jaster, P., & Thill, C. (2002b). Plant responses under an experimental near infra-red reflecting greenhouse film. Acta Horticulturae, 580, 137e143. Sadler, E. J., & Van Bavel, C. H. M. (1984). Simulation and measurement of energy partition in a fluid-roof greenhouse. Agricultural and Forest Meteorology, 33(1), 1e13. Sonneveld, P. J., Swinkels, G. L. A. M., Bot, G. P. A., & Flamand, G. (2010). Feasibility study for combining cooling and high grade energy production in a solar greenhouse. Biosystems Engineering, 105(1), 1e58. Stanghellini, C., & De Jong, T. (1995). A model of humidity and its applications in a greenhouse. Agricultural and Forest Meteorology, 76(2), 129e148. Stanhill, G., Fuchs, M., Bakker, J., & Moreshet, S. (1973). The radiation balance of a glasshouse rose crop. Agricultural Meteorology, 11, 385e404. Tanaka, J. (1997). A model experiment on reduction of greenhouse cooling load in the daytime using shading material against infrared radiation as covering. Environment Control in Biology, 35, 15e20. Van Bavel, C. H. M., Damagnez, J., & Sadler, E. J. (1981). The fluidroof solar greenhouse: energy budget analysis by simulation. Agricultural Meteorology, 23(1), 61e76. Van Rijssel, E. (2006). Nieuw krijtmiddel neemt bij regen fors minder licht weg: nog krijt in 2006 of al een modern schermmiddel? Onder Glas, 3(2), 54e55. Van Staalduinen, J. (2008). In combinatie met dekberegening werkt ReduHeat prima. Onder glas, 5(4), 68e69. Vegter, B. (2007). Wikken en wegen bij krijten. Vakblad voor de bloemisterij, 62(10), 34e35. Verlodt, I., & Verschaeren, P. (1997). New interference film for climate control. Plasticulture, 115, 27e35. Von Elsner, B., Xie, J., 2003. Effects of interference pigments in shading paint for greenhouses. Proceedings of the thirty-first agricultural plastics congress, August 16e19, 2003 in Grand Rapids, Michigan, USA, 6e16.