Abstract
Water management is one of the most important operations in greenhouse baby leaf production. However, growers mainly irrigate the plants based on experience, which generally leads to yield loss, uneven quality, and low water-use efficiency. This study evaluated four evapotranspiration (ET) models, such as Radsum, Penman methods, FAO Penman-Monteith, and Priestley-Taylor, for irrigation strategy by predicting the ET level of greenhouse baby pakchoi [Brassica rapa L. ssp. chinensis (L.) Hanelt] under different plant densities (72-, 128-, 200-, and 288-plug tray). Among environmental factors, net radiation and photosynthetically active radiation (PAR) had the highest correlation with ET, with R2 of 0.93 and 0.94, respectively. Plant growth period was divided into different stages according to canopy development and substrate surface coverage. The corresponding crop coefficient (Kc) was introduced into ET prediction models. The result shows overestimation of ETc (crop evapotranspiration) by the Radsum and Penman methods. FAO Penman-Monteith and Priestley-Taylor methods performed the best with R2 ≈0.7 for all planting densities. These two methods are recommended for greenhouse irrigation scheduling in baby pakchoi production.
With growing interest in local food production using greenhouse technology, fresh and nutritious baby leaf greens are becoming increasingly popular among consumers (Kroggel et al., 2012). Plug trays filled with substrate is used for greenhouse baby leaf production with ebb-and-flow (E&F) irrigation systems (Danfeng et al., 2013). E&F irrigation is characterized as producing uniform plants, with a short production cycle, high water- and fertilizer-use efficiency, and adequate plant quality (Yang et al., 2018). In production practice, the quality of baby leaf is sensitive to substrate water content. Water deficiency or overirrigation can have a significantly negative effects on baby leaf plant physiology (Guo et al., 2017). Therefore, it is critical to manage irrigation precisely in greenhouse baby leaf production to improve production quality and water-use efficiency.
Greenhouse growers typically irrigate crops based on their personal experience, which normally leads to uneven product quality and waste of water resources. Few studies have reported the irrigation scheduling strategy for leafy greens by using subirrigation such as E&F irrigation. Irrigation frequency and flood time were identified as two key factors in the E&F irrigation strategy (Yang et al., 2018). In substrate type, flooding depth and time have also been studied to optimize E&F irrigation (Anlauf et al., 2012).
Different methods have been developed for greenhouse irrigation management. The main approaches include soil-based, plant-based, and climate-based methods (Jones, 2004). The monitoring of water content or water potential in growing media is a fast and reliable way to establish irrigation scheduling. However, sampling position and sensor cost limit the application of this method in baby leaf production in trays. Weighting the tray could also accurately measure the water amount, but this method has high labor costs. The plant phenotyping technique can be used to determine plant response to different water content with cameras scanning the plants (Guo et al., 2017). In this way, irrigation time could be indicated by plant phenotypes, but the irrigation amount is difficult to quantify.
Irrigation water requirements can be defined as the quantity of irrigation water required to produce the desired crop yield and quality and to maintain an acceptable water balance in the substrate. Previous studies showed that water requirement for different crops was highly related to the light radiation, vapor pressure deficit (VPD), substrate characteristics, and crop growth characteristics in greenhouse (Orgaz et al., 2005; Sumner and Jacobs, 2005). Predicting ET is a commonly used indicator of irrigation decision-making, which is the main index for determining the optimal crop water and fertilizer management (Allen et al., 1998).
In greenhouses, the soil, crop, and environment are regarded as a continuum to study the water absorption through crop roots, the transport of water in crops, and the transport of water inside plants (Campbell and Norman, 2000). In this process, more attention should be paid to crop response to irrigation and greenhouse microclimate, as well as the combination of agronomy, plant physiology, environmental control, engineering technology, and other factors. Optimized ET models and greenhouse thermal environment management enhance greenhouse energy conservation, reduce water consumption, and improve crop quality and yield. Several studies have investigated ET prediction in greenhouse. The model varies across different regions, climates, and greenhouse environments; hence, the application is often limited. Whether it is applicable in other regions or other greenhouses, systemic calibration and validation should be replicated in more diverse greenhouse climates.
The use of the ET model in a greenhouse environment is different from the open field because of the variation and dynamic climate. Existing studies about ET prediction in greenhouse mainly focus on vegetables such as tomato, cucumber, and eggplants. Seasonal ET of melon, green beans, watermelon, and pepper was studied in an unheated plastic greenhouse (Orgaz et al., 2005). Three models (Stanghellini, Penman-Monteith, and Takakura) were compared to simulate the ET of bell pepper and tomato. Results showed that the Stanghellini model has the best overall performance under all conditions evaluated, but there were no significant differences between the three models. (Villarreal-Guerrero et al., 2012). Several models developed to estimate ET in greenhouses have been reviewed, suggesting greenhouse ET models should be selected depending on the type, location, and crop type (Karaca et al., 2018). In another review article, 10 models used in different type of greenhouses were compared, and the FAO Penman model was recommended for plastic greenhouse ET estimation (Ilahi, 2009). To date, limited studies have been done on ET research on baby leaf vegetables in greenhouse production.
There have been several investigations into the ET modeling focused on leafy greens or short plants in greenhouse. Floating seedling water evaporation in greenhouse was investigated using an ET model based on the Penman formula under artificial lighting conditions (Zeng et al., 2015). A canopy transpiration model for greenhouses was established based on the Penman-Monteith transpiration model to simulate the ET of pakchoi in a plastic greenhouse covered with insect proof net (Jun et al., 2009). ET of turfgrass in greenhouse was studied with the FAO-56 Penman-Monteith and a 20-cm evaporating pan method with adequate correlation between actual ET (ETa) and reference ET (ET0) predicted by the FAO Penman-Monteith model (Xue et al., 2008). Andersson evaporimeters, FAO Radiation, and FAO Penman-Monteith equations were compared with estimate lettuce ET in greenhouse (Casanova et al., 2009). The Priestley-Taylor model was evaluated for estimating greenhouse tomato with drip irrigation, but no research was found for greenhouse leaf vegetables (Valdés-Gómez et al., 2009). FAO-24 Penman and Priestley-Taylor were compared with FAO-56 Penman-Monteith in Venlo-type greenhouse ET prediction with the conclusion that the application of FAO-56 Penman-Monteith in the greenhouse was unsatisfactory and modification was needed to adjust the Aerodynamic resistance (Eitzinger et al., 2007). The Penman-Monteith and Priestley-Taylor models can directly estimate ETa but require canopy resistance. These variables depend on leaf area index, soil water availability, elevation, soil characteristics, VPD, and solar radiation (Sumner and Jacobs, 2005). Water management of greenhouse tray seedling leafy vegetables lacks a basis for decision-making, relying mostly on empirical selection. ET model studies rarely involve greenhouse seedling leafy vegetables.
The objectives of this study were 1) to assess the applicability of four ET prediction methods in greenhouse baby pakchoi production; and 2) to investigate the performance of the ET prediction models on baby pakchoi irrigation management under different plant densities. Climate-based methods could provide a potential solution for greenhouse smart irrigation scheduling.
Materials and Methods
Experiment condition and greenhouse management
The greenhouse experiments were conducted from 1 Sept. to 23 Sept. on baby pakchoi, cv. Huawang. The seedlings were grown in a multispan plastic greenhouse located at Jiading, Shanghai (lat. 31.36°N, long. 121.21°E, 17 m altitude). The experimental greenhouse was a GSW-8430 type four span greenhouse with single layer polyethylene film. The greenhouse had total area of 3974 m2, with top height of 5.2 m, span of 8 m, and gutter height of 3 m. The greenhouse was equipped with 70% internal shading screen, electrical top windows, manual side windows, E&F seedbed, and irrigation control system. The greenhouse climate was controlled by manual ventilation and shading. During the experiment, manual ventilation was used several times per day according to the climate conditions. The side window was kept open during the experiment period (late summer in Shanghai) to maintain the natural ventilation. The top window was opened when the air temperature was >25 °C and closed at night and on cloudy days. E&F irrigation operated twice per day at 0800 hr and 1600 hr, and each irrigation lasted ≈7 min (flood time). In the preexperiment, three irrigation frequencies were compared, and the result showed that twice daily irrigation could fully saturate the substrate. During this experiment, the baby pakchoi were grown in four types of tray (72-, 128-, 200-, and 288-plug trays, tray size 54 × 28 cm). Each group had 21 trays and three random group replications.
The pakchoi was sown into plug trays filled with commercial seedling substrate (Jinhai Agriculture Technology Inc., Hangzhou, China) on 1 Sept. After germinating for 72 h under 25 °C and 95% relative humidity (RH) in the germination room, trays with seedlings were transferred to the greenhouse and irrigated with nutrient solution (EC = 0.5 mS/cm and pH 5.5 to 6.5). Trays were placed on movable E&F benches, 0.7 m aboveground.
Variables measurement
The PAR, air temperature, and RH were measured by sensors with a HOBO 3.0 climate station (Onset, Bourne, MA) located 2 m aboveground. Net radiation was measured by four-component net radiometer (NR01; Hukseflux, the Netherlands) 2.5 m above ground level. All sensors were scanned at 5-s intervals, and 5-min averaged readings were recorded. Plant biomass was obtained every 2 d until baby pakchoi was harvested on 23 Sept. Average daily actual ET of seven trays from different treatments (72-, 128-, 200-, and 288-plug trays) was measured as ETa according to the water balance method (Yuan et al., 2001). ETa was measured as the mean weight differences of seven trays before and after each irrigation at 0800 hr and 1600 hr with electronic scale (Scale: 0 to 6 kg, Accuracy: 0.1 g, D11; Sanfeng, Shanghai, China). Leaf area was scanned at 2-d intervals by using a Canon digital scanner (Canonscan LiDE 120; Canon Inc., Tokyo, Japan), and the images were analyzed with ImageJ to extract the leaf area data (Schneider et al., 2012). The total leaf area per plant from different groups were obtained to get the leaf area index (LAI). LAI was calculated as the ratio of one-sided leaf area to the total tray area (54 × 28 cm). Growth rate of LAI was calculated by the percentage increase at 2-d intervals during the experiment.
Evapotranspiration model description
The application of ET model relies on the availability of data, which is limited by the measurement equipment, expertise, climate, and historical data. Therefore, models reported in previous studies with fewer inputs and better greenhouse performance were selected (Ilahi, 2009). Two physical ET models (Penman model and FAO Penman-Monteith model) and two radiation-based models (Priestley-Taylor model and simplified radiation model) were evaluated in this study for their applicability in greenhouse leaf vegetable irrigation practice. Detailed information on the models is provided in Table 1, together with description and value set in this study of variables and constants required by the four models.
Variables and constants used in these evapotranspiration (ET) models.
The Penman method.
The FAO Penman-Monteith model.
Priestley-Taylor model (PT, 1972).
Simplified radiation sum model (Radsum, 2016).
Crop type and growth stages are the major factors influencing the value of Kc, which is strongly related to the substrate surface area covered by the crop canopy and to LAI.
Evaluation of model performance
The R2 and Akaike information criterion (AIC) were used for comparing the models. The R2 value measures how well the model results fit the actual value, with the value close to 1 indicating more variance explained by the model. AIC is a technique based on in-sample fitness to estimate the likelihood of a model, which is used further to estimate the predicted values. A good model is one that has the minimum AIC compared with all the other models (Akaike, 1974).
The average daily ETc were calculated by the four models with input variables including daily environment data, such as temperature, radiation, humidity, and wind speed. R language program software with the Evapotranspiration package was used for such calculations (Guo et al., 2016).
where n is the number of observations, Pi is the estimated ETc by different models (predicted value), and Oi is the measured ETa (the actual daily ET). Statistical analysis was conducted in R 3.4.2 and ET model were simulated using the Evapotranspiration package (Guo et al., 2016).
Results
Climate conditions and plant growth under different plant densities.
Figure 1 shows the daily average climate data during the experimental period. The net radiation level was the highest in sunny days with the daily average value >80 W·m−2. The net radiation in cloudy days was less than 50 W·m−2. The corresponding ETa was 2.7 mm·d−1 on sunny days and 0.7 mm·d−1 on cloudy days. RH varied from 65% to 95% during the experiment and tended to change in opposition to radiation and air temperature. The corresponding RH is 75% on sunny days and 90% on cloudy days. Average daily air temperature varied from 21 to 29 °C. The corresponding daily air temperature was 25.7 °C in sunny days and 24.1 °C in cloudy days. As air temperature increases, air can hold more water, and RH decreases. ET is driven by VPD, which is a function of temperature and RH. As VPD increases, the plant transpires faster due to the larger difference in vapor pressures between the leaf and the air. There was a significant correlation between weather data and ETa, indicated by the coincident occurrence of peaks and valleys (Fig. 1).
Greenhouse climate factors and measured actual average daily evapotranspiration (ETa). PAR = photosynthetically active radiation.
Citation: HortScience horts 56, 2; 10.21273/HORTSCI15513-20
The Pearson correlation analysis was conducted to get a better understanding of the relationship between ETa and climate data. As shown in Table 2, a positive correlation was found between ETa and radiation/PAR (r > 0.9, P < 0.001) and a negative correlation between ETa and RH (r = –0.77, P < 0.001), whereas the correlation was not significant between ETa and daily air temperature with r = 0.46 and P > 0.05.
Correlation analysis of climate factors with ETa.
The LAI in different tray size at harvest stage varied from 3.3 to 7.0. LAI increased with increasing density. As Fig. 2A shows, the 288-plug tray maintained the highest LAI throughout the entire period, whereas the 72-plug tray had the lowest due to the density difference. LAI of baby pakchoi in 288-plug tray reached 3.0 at 14 d after sowing (DAS), whereas LAI of the 72-plug tray reached 3.0 at 20 DAS. The LAI growth rate (compared with the day before) reached a peak value (206% for 72-plug tray and 160% for 288-plug tray) at 8 to 10 DAS (Fig. 2B). The first true leaf was expanded in 8 to 10 DAS compared with former stages, which made the peak value of LAI growth rate. After 10 DAS, the growth rate of LAI decreased. The increase rate of LAI at different densities showed a different trend from that of LAI value. The highest increase rate occurred in 72-plug tray (Fig. 2B).
Dynamic of (A) leaf area index (LAI) and (B) LAI growth rate of baby pakchoi in different plant densities (72, 128, 200, and 288 plants/tray). Dashed line indicates the tray was fully covered by plant canopy. DAS = days after sowing.
Citation: HortScience horts 56, 2; 10.21273/HORTSCI15513-20
Determination of crop coefficient.
According to FAO-56, Kc is mainly affected by the crop characteristics, agronomy, growth stages, and climate conditions. The frequency of rain or irrigation is important during the early growth stage. The Kc values can be classified into four stages: initial, crop development, midseason, and late season. In this research, the baby leaf grows only for the first three stages before harvest. The initial period represents the period with 10% groundcovered by crops, and the development period extends from the end of the initial stage to the effective full cover (Allen et al., 1998). Midseason extends from effective full cover to when plants begin to decrease. The effective full cover can be predicted when the crop reaches a LAI of 3.0. For baby leafy vegetable, taking baby pakchoi as the example, the initial stage starts from sowing to the cotyledons fully expanded with coverage under 10%. In this stage, substrate surface evaporation contributes the most to the ET. The development starts from emergence of the first true leaf to full coverage with LAI = 3.0. These stages are highly related to plant density and growth rate.
Soil evaporation in the initial stage contributes the most to crop water demand. Therefore, the impact of precipitation or irrigation needs to be considered when determining the Kc ini value. Kc ini is related to the ET0 and interval between irrigations. As the irrigation frequency in greenhouse was twice per day, the substrate surface kept being wet. According to the relation curve of average Kc ini and ET0 and the interval between irrigations >40 mm per event for medium and fine textured soils in FAO 56, when the ET0 is <5 mm·d−1 and the irrigation event occurs less than 1 d, the Kc ini value should be 1.15. The Kc mid is recommended as 1.05 for small vegetables at the full growth stage (with LAI >3.0) (Allen et al., 1998). Table 3 lists the growth stage and the LAI of different plant densities. The time interval was decided by LAI for each group. The Kc values of each stage under different plant densities are listed correspondingly.
Crop coefficient (Kc) at different stages in this study.
Estimation of crop ETc.
By introducing Kc to the four evapotranspiration models, we obtained the predicted daily ETc of the entire 17 d of experiment period (Fig. 3). During the experiment period, the fluctuation pattern of predicted ETc was similar to the measured ETa.
Comparison of different evapotranspiration (ET) models in predicting the daily crop ET (ETc) vs. actual average daily ET (ETa). Radsum = simplified radiation sum model; FAO-PM = FAO Penman-Monteith model; PT = Priestley-Taylor model.
Citation: HortScience horts 56, 2; 10.21273/HORTSCI15513-20
The FAO-PM and PT models had higher accuracies in predicting ETc compared with the Radsum and Penman models, which overestimated ETc under most conditions (Fig. 3). The FAO-PM and PT models showed better performance on sunny day than cloudy days. Overestimates were found in cloudy day such as DAS of 6, 7, 10, and 11. MAPE, RMSE, R2, and AIC were calculated to indicate the performance of the four models in different plant densities (Table 4).
Evaluation of model performance under different plant densities.
The overall performance showed that FAO-PM and PT models performed better than the other two models in different plant density groups. The performance of Penman model ranked the last. MAPE <50% was considered appropriate model performance. Hence, PT, FAO-PM in 200-plug tray group, and Radsum in the first three groups (72-, 128-, and 200-plug trays) were acceptable. Models with smaller RSME represent better prediction of results. FAO-PM and PT models showed better results (RMSE = 0.4 to 0.6 mm·d−1) than Radsum and Penman models in four groups. Meanwhile, the FAO-PM and PT models had a higher R2 value at ≈0.7 and lower a AIC value than those of Radsum and Penman in different plant density groups.
Discussion
ET is affected by variations in greenhouse environmental factors, and the order of correlation with ETa is PAR > net radiation > humidity > VPD > air temperature. This result is consistent with those of Orgaz and Casanova, who also concluded that solar radiation was the main factor affecting ET (Casanova et al., 2009; Orgaz et al., 2005).
The dynamic change of LAI was significantly different among the four planting densities groups in our study, but the difference of measured ETa was not statistically significant. Some studies reported that both Kc and ETc were highly correlated with LAI (Netzer et al., 2009), and higher LAI increased ET as biomass development related to water use efficiency (Jha et al., 2018). In our study, the main reason that LAI had less effect on ET may be the short growth period of baby leaf having less impact of LAI on Kc. Also, the relatively high-frequency irrigation kept the substrate surface wet, and the evaporation from the surface of the substrate was relatively higher than plant transpiration on the growth stage.
The predictions of the four models show that the Penman and Radsum models did not predict the actual evapotranspiration well. In contrast, FAO-PM and PT provided better predictions, with R2 values of ≈0.7. The Radsum model only considered the radiation as input, without humidity, for the ET prediction. In the case of extreme humidity conditions, the prediction results could be biased (Sentelhas et al., 2010).
The FAO-PM model, a function of net radiation and VPD, was less effective compared with the PT model, which agrees with Sumner’s observations (Sumner and Jacobs, 2005). The FAO-PM method would seriously underestimate the ET0 in a greenhouse environment with small wind speed (Fernandez et al., 2010). In this research, the average wind speed above the plant canopy in greenhouse was ≈1 m·s−1 with circulating-fan ventilation. Therefore, the FAO-PM model is applicable for the well-ventilated greenhouse. The Penman model mainly considers the energy balance and does not take into account the dynamic changes of plants, so the prediction accuracy was lower than FAO-PM. Compared with the FAO-PM model, the PT model had fewer parameters. It is more suitable for the conditions of the uniform wet ground surface, which fits the conditions of greenhouse used (Sumner and Jacobs, 2005). After adjustment by Kc, the PT model is recommended for the prediction of greenhouse evapotranspiration. These models can be used for daily irrigation scheduling by being implemented in a decision support system. The calibration of parameters such as crop resistances and αpt (i.e., Priestly Taylor) may improve the prediction results (Villarreal-Guerrero et al., 2012).
Higher accuracy of FAO-PM and PT than Radsum and Penman demonstrating the importance of the VPD portion of these models (Prenger et al., 2002). With more comprehensive environmental variables, the models have better prediction efficiency. There are uncertainties associated with different models due to data and structure. A different model should be developed to minimize the biases in the further research.
Conclusion
In this study, four ET models were implemented in greenhouse baby pakchoi irrigation scheduling, and the performances of these models were evaluated. The ETc (crop evapotranspiration) prediction results proved that the FAO-PM and PT models performed better than the Penman and Radsum models in greenhouse. This model-based precision irrigation scheduling portends to be meaningful in a decision support system to guarantee the water requirements of greenhouse baby pakchoi tray production.
Literature Cited
Akaike, H. 1974 A new look at the statistical model identification ITAC 19 716 723 doi: 10.1109/TAC.1974.1100705
Allen, R.G., Pereira, L.S., Raes, D. & Smith, M. 1998 Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56. FAO, Rome 300:D05109
Anlauf, R., Rehrmann, P. & Schacht, H. 2012 Simulation of water uptake and redistribution in growing media during ebb-and-flow irrigation J. Hort. For. 4 8 21 doi: 10.5897/JHF11.058
Campbell, G.S. & Norman, J.M. 2000 An introduction to environmental biophysics. Springer, New York
Casanova, M., Messing, I., Joel, A. & Canete, A. 2009 Methods to estimate lettuce evapotranspiration in greenhouse conditions in the central zone of Chile Chil. J. Agr. Res. 69 60 70 doi: 10.4067/S0718-58392009000100008
Danfeng, H., Leiling, S. & Kai, Z. 2013 Primary study on pakchoi factory Acta Hort. 79 84 doi: 10.17660/ActaHortic.2013.1004.7
Eitzinger, J., Schahbazian, N., Allahdadi, I. & Akbari, G. 2007 A decision support system for the optimization irrigation scheduling Pakistan J. Water Resources 11 1 18
Fernandez, M.D., Bonachela, S., Orgaz, F., Thompson, R., Lopez, J.C., Granados, M.R., Gallardo, M. & Fereres, E. 2010 Measurement and estimation of plastic greenhouse reference evapotranspiration in a Mediterranean climate Irr. Sci. 28 497 509 doi: 10.1007/s00271-010-0210-z
Guo, J., Juan, L., Chang, L., Zhang, J. & Huang, D. 2017 Discrimination of plant root zone water status in greenhouse production based on phenotyping and machine learning techniques Sci. Rep. 7 8303 doi: 10.1038/s41598-017-08235-z
Guo, D.L., Westra, S. & Maier, H.R. 2016 An R package for modelling actual, potential and reference evapotranspiration Environ. Model. Softw. 78 216 224 doi: 10.1016/j.envsoft.2015.12.019
Ilahi, W.F.F. 2009 Evapotranspiration models in greenhouse. Wageningen University, Netherlands, M.Sc
Jha, P.K., Kumar, S.N. & Ines, A.V.M. 2018 Responses of soybean to water stress and supplemental irrigation in upper Indo-Gangetic plain: Field experiment and modeling approach Field Crops Res. 219 76 86 doi: 10.1016/j.fcr.2018.01.029
Jones, H.G. 2004 Irrigation scheduling: Advantages and pitfalls of plant-based methods J. Expt. Bot. 55 2427 2436 doi: 10.1093/jxb/erh213
Jun, L.I., Yao, Y.P., Luo, W.H., Dai, J.F., Ning, H.U., Rui, X.U., Han, X.B. & Zhang, H.J. 2009 Transpiration of Brassica chinensis L. in a plastic greenhouse covered with insect-proof nets in lower reaches of Yangtze River: A simulation study Chinese J. Appl. Ecol. 20 2241 2248
Karaca, C., Tezcan, A., Buyuktas, K., Buyuktas, D. & Bastug, R. 2018 Equations developed to estimate evapotranspiration in greenhouses Yüz. Yil Üniv. Tarim Bilim. Derg. 28 482 489 doi: 10.29133/yyutbd.427115
Kroggel, M., Lovichit, W., Kubota, C. & Thomson, C. 2012 Greenhouse baby leaf production of lettuce and komatsuna in semi-arid climate: Seasonal effects on yield and quality Acta Hort. 952 827 834
Netzer, Y., Yao, C., Shenker, M., Bravdo, B.-A. & Schwartz, A. 2009 Water use and the development of seasonal crop coefficients for Superior Seedless grapevines trained to an open-gable trellis system Irr. Sci. 27 109 120 doi: 10.1007/s00271-008-0124-1
Orgaz, F., Fernandez, M.D., Bonachela, S., Gallardo, M. & Fereres, E. 2005 Evapotranspiration of horticultural crops in an unheated plastic greenhouse Agr. Water Mgt. 72 81 96 doi: 10.1016/j.agwat.2004.09.010
Penman, H.L. & Keen, B.A. 1948 Natural evaporation from open water, bare soil and grass Proc. R. Soc. Lond. A Math. Phys. Sci. 193 120 145 doi: 10.1098/rspa.1948.0037
Prenger, J.J., Fynn, R.P. & Hansen, R.C. 2002 A comparison of four evapotranspiration models in a greenhouse environment Trans. ASAE 45 1779 doi: 10.13031/2013.11429
Priestley, C. & Taylor, R. 1972 On the assessment of surface heat flux and evaporation using large-scale parameters MWRv 100 81 92
Schneider, C.A., Rasband, W.S. & Eliceiri, K.W. 2012 NIH Image to ImageJ: 25 years of image analysis Nat. Methods 9 671 675 doi: 10.1038/nmeth.2089
Sentelhas, P.C., Gillespie, T.J. & Santos, E.A. 2010 Evaluation of FAO Penman–Monteith and alternative methods for estimating reference evapotranspiration with missing data in Southern Ontario, Canada Agr. Water Mgt. 97 635 644 doi: 10.1016/j.agwat.2009.12.001
Sumner, D.M. & Jacobs, J.M. 2005 Utility of Penman-Monteith, Priestley-Taylor, reference evapotranspiration, and pan evaporation methods to estimate pasture evapotranspiration JHyd 308 81 104 doi: 10.1016/j.jhydrol.2004.10.023
Valdés-Gómez, H., Ortega-Farías, S. & Argote, M. 2009 Evaluation of the water requirements for a greenhouse tomato crop using the Priestley-Taylor method Chil. J. Agr. Res 69 3 11 doi: 10.4067/s0718-58392009000100001
Villarreal-Guerrero, F., Kacira, M., Fitz-Rodriguez, E., Kubota, C., Giacomelli, G.A., Linker, R. & Arbel, A. 2012 Comparison of three evapotranspiration models for a greenhouse cooling strategy with natural ventilation and variable high pressure fogging Scientia Hort. 134 210 221 doi: 10.1016/j.scienta.2011.10.016
Waller, P. & Yitayew, M. 2016 Evapotranspiration, p. 67–87. In: Irrigation and drainage engineering. Springer International Publishing, Cham
Xue, L., Hanping, M., Zhiyu, Z., Xiaodong, Z. & Weiguo, F. 2008 Experimental study on prediction of turf transpiration in greenhouse. Agricultural Mechanization Res. 12:119–121, doi: 10.3969/j.issn.1003-188X.2008.12.037
Yang, L.Q., Yang, X., Zhao, H., Huang, D.F. & Tang, D.Q. 2018 Ebb-and-flow subirrigation strategies increase biomass and nutrient contents and reduce nitrate levels in lettuce HortScience 53 1056 1061 doi: 10.21273/Hortsci13065-18
Yuan, B.-Z., Kang, Y. & Nishiyama, S. 2001 Drip irrigation scheduling for tomatoes in unheated greenhouses Irr. Sci 20 149 154 doi: 10.1007/s002710100039
Zeng, C., Yang, X., Fan, C. & Wu, M. 2015 Preliminary study on characteristics of water evaporation of floating seedling greenhouse Chinese Agr. Sci. Bul. 31 262 266
