METHODOLOGY
-----------

Greenhouse data collection
--------------------------

Data was collected at WUR Greenhouse Horticulture, Violierenweg 1, 2665 MV Bleiswijk, The Netherlands, between 19 October 2009 at 15:15 and 8 February 2010 at 15:15.

Tomato plants (variety Sunstream) were grown using a high wire cultivation system, transplanted to the greenhouse on 16 October 2009. Two neighbouring compartments within an experimental greenhouse, one equipped with HPS top-lights (electric input of 110 W/m2, with an efficacy of 1.8 µmol PAR/J), and one with LED top-lights (electric input of 116 W/m2, with an efficacy of 1.6 µmol PAR/J and an active water cooling system). Both compartments had a pipe rail and grow pipe heating system, thermal and blackout screens. The size of each compartment was 9.6 m by 15 m, with a total floor area of 144 m2, eave height of 5.7 m and ridge height of 6.7 m. The roof of each compartment consisted of 2 ridges, with a slope of 23°. The compartments were part of a larger greenhouse measuring 120 m by 80 m. Two walls of each compartment faced neighbouring compartments, two walls faced indoor corridors, and the roof faced the outdoor. 

The compartments were each equipped with a pipe rail system (a heating system of pipes laid out on the greenhouse floor, also used as rails for trollies) and a grow pipe system (a heating system of pipes hung at crop height). 

The grow pipe system in each compartment consisted of 7 pipes, each at a diameter of 35 mm and length of 13.5 m. In addition, each of the 7 pipes was hung by 2 hoses, each 3 m long and 29 mm wide, and connected to 2 connectors, each 2 m long and 29 mm wide. An additional section of the grow pipe system was installed on the sidewalls, with a total length of 9.4 m and a diameter of 64 mm. Lastly, a section of the system was installed on the back wall, with a total length of 37 m and a diameter of 51 mm.

In total, the grow pipe system in each compartment consisted of 94.5 m of pipes at a diameter of 35 mm; 70 m of pipes at a diameter of 29 mm; 9.4 m of pipes at a diameter of 64 mm; and 37 m of pipes at a diameter of 51 mm.

The pipe rail system in each compartment consisted of 6 rails, with a total length of 166.8 m and a diameter of 51 mm. Each rail was connected by 2 hoses, 0.5 m long and 29 mm wide, to a total length of 6 m. Furthermore, the network consisted of 3 pipes, each 9.4 m long and 58 mm wide, with a total length of 19.2 m.

In total, the pipe rail system in each compartment consisted of 166.8 m of pipes at a diameter of 51 mm; 6 m of pipes at a diameter of 29 mm; and 19.2 m of pipes at a diameter of 58 mm.

The maximal CO2 injection rate in each compartment was 720 mg/s per compartment.

In each compartment, 12 plant rows were sown, with a plant density and initial stem density of 3.12 plants and stems per m2. The stem density was increased to 3.9 stems per m2 on December 14, 2009, and to 4.7 stems per m2 on January 27, 2010. In the HPS compartment, HPS lamps were hung above two paths, 8 lamps of 1000 W above each path. Lamps were installed at a height of 4.7 m and with a distance of 1.85 m between the lamps. 

In the LED compartment, lamps were installed above the crop rows. LEDs from Lemnis Lighting, the Netherlands were used. These lamps were water-cooled to maintain their efficacy, and heat extracted by the cooling system was removed from the greenhouse. The LED lighting was composed of 12% blue LEDs (with a peak at 450 nm) and 88% red LEDs (with a peak at 660 nm). The LEDs were installed at a height of 4.65 m. 

In both compartments, the light distribution from the lamps was measured during the night using a Sunscan Canopy analysis system (Delta-T Ltd, Cambridge, UK), to ensure a uniform distribution of PAR light from the lamps. The PPFD from the lamps above the crop was 170 µmol/m2/s in both compartments. The maximum daylength was 18 h, and the lamps were switched off one hour before sundown. The setting for CO2 concentration was 1000 ppm. Irrigation, leaf and flower pruning, and temperature set points were modified dynamically by observing the state of the crop with a team of experts with the aim of maximizing production. 

Plants in the HPS compartment were slightly taller than those in the LED compartment throughout the trial. Similarly, the leaf area index (LAI, leaf area per floor area, m2/m2), was consistently higher in the HPS compartment, and to a lesser extent, the average number of trusses per stem. Further information is given in: 
	T Dueck, J Janse, A Schpendonk, F Kempkes, B Eveleens, K Scheffers, S Pot, G Trouwborst, E Nederhoff, L Marcelis. Lichtbenuttig van tomaat onder LED en SON-T belichting. November 2010. Rappport GTB-1040. Wageningen UR Glastuinbouw, Wageningen, The Netherlands. https://research.wur.nl/en/publications/lichtbenutting-van-tomaat-onder-led-en-son-t-belichting-2, https://edepot.wur.nl/160925 (In Dutch)
and:
	TA Dueck, J Janse, BA Eveleens, FLK Kempkes, LFM Marcelis. Growth of tomatoes under hybrid LED and HPS lighting. 2012. Acta Horticulturae, 1(952), 335–342. https://doi.org/10.17660/ActaHortic.2012.952.42
and in:
	D Katzin, S van Mourik, F Kempkes, EJ van Henten. GreenLight: An open source model for greenhouses with supplemental lighting: Evaluation of heat requirements under LED and HPS lamps. 2020. Biosystems Engineering 194, pp. 61-81. https://doi.org/10.1016/j.biosystemseng.2020.03.010

The data in the processed data files is organized in the following way:

Column  Description                         Unit             
1 		Time 								MATLAB datenum (days since 0/0/0000)
2 		Outdoor global radiation 			W m^{-2}
3 		Outdoor air temperature 			°C
4 		Outdoor relative humidity  			%		
5 		Outdoor wind speed 					m s^{-1}
6 		Indoor temperature 					°C
7 		Indoor vapor pressure deficit  		g m^{-3}
8 		Energy screen closure 				0-100 (100 is fully closed)
9 		Black out screen closure			0-100 (100 is fully closed)
10 		Lee side ventilation aperture		0-100 (100 is fully open)
11 		Wind side ventilation aperture 		0-100 (100 is fully open)
12 		Pipe rail inlet temperature 		°C, except values of 0 indicate the pump was off (no heating was actively provided)
13 		Grow pipes inlet temperature 		°C, except values of 0 indicate the pump was off (no heating was actively provided)
14 		Lights on/off 						0/1 (1 is on)
23		CO2 injection status				0/1 (1 is on)
24 		Indoor CO2 concentration 			ppm

Time (column 1) was converted from Excel format to MATLAB format by using the MATLAB datetime function.

The outdoor weather data (files "Weather raw.csv" and "Weather raw.xlsx") was measured using a Hoogendoorn weather mast (Hoogendoorn, Vlaardingen, The Netherlands) located on top of the greenhouse service building. This data was used for columns 2-5 in the processed data.

The indoor climate data was measured by a ventilated Hoogendoorn measurement box, which was placed in the middle of the compartment at the height of the top of the crop. The height of the measurement box was adjusted throughout the experiment to keep it at the top of the crop. 

Climate actuators were controlled by an Economic Hoogendoorn climate control. Indoor climate data and climate control data (files "LED raw.csv", "LED raw.xlsx", "HPS raw.csv", "HPS raw.xlsx") was collected in 5 min intervals by an Economic Hoogendoorn climate controller. This data was used for columns 6-14 and 23-24 in the processed data.

CO2 injection data (column 23 above) was recorded according to momentary information at 5 minute intervals. Note that this means that a value of 1 for this column at any given timepoint indicates that CO2 injection was supplied (set to a rate of 720 mg/s per compartment) at the moment of recording, and not neccesarily throughout the entire 5 minutes between two moments of recording.

Columns 6, 8-14, and 23-24 in the processed data were based on the raw data measurements, with missing values in the raw data measurements filled by linear interpolation for the processed data. Column 7 in the processed data (indoor vapor pressure deficit) was calculated using the function rh2vaporDens, available at:
	https://github.com/davkat1/GreenLight/blob/master/Code/serviceFunctions/rh2vaporDens.m
	
The indoor temperature and relative humidity raw data was converted to indoor vapor pressure deficit by using:
	vpd = 1e3*(rh2vaporDens(temp,100)-rh2vaporDens(temp,rh));
	
where temp is the indoor temperature (°C), rh is the indoor relative humidity (%), and vpd is the calculated vapor pressure deficit (g m^{-3}).

Apparent sky temperature, used in simulations, was calculated based on the measured outdoor air temperature data (column 3 as described above) and cloud cover data recorded by KNMI (Royal Netherlands Meteorological Institure) in Rotterdam. See https://github.com/davkat1/GreenLight/blob/master/Code/inputs/skyTempRdam.m and https://github.com/davkat1/GreenLight/tree/master/Code/inputs/cloudCoverRotterdam2009-2012 for more information.

Simulated data
--------------

The code in "Simulation data\Source code", subfolders "BramVanthoorModelVer2" and "StateSpaceModel" was used to generate the data in the original Katzin et al. 2020 paper. The code was run using MATLAB R2018b. The results of those simulations are in "Simulation data\MATLAB output\Original data 2019".

In order to test that this is indeed the correct version of the code, simulations were run again in July 2023. It turned out that the MATLAB version used to run the simulation influences the final result. The changes are very small, and likely due to numerical differences in how the built-in MATLAB ODE solver behaves. In any case, the scripts "Source code\runKatzin2020simulations.m" and "Source code\generateKatzin2020FiguresTables.m" were written in July 2023 by David Katzin, in order to reproduce the data in the original paper. The results of these simulations (performed in July 2023, using MATLAB R2018b on Windows 10) are in "Simulation data\MATLAB output\Replicated 2023".

Finally, the simulation results in "Simulation data\MATLAB output\Original data 2019" were converted to CSV files by using the script generateCsvData.m (performed in August 2023 by David Katzin, using MATLAB R2021b on Windows 10). The results of this script are in "Simulation data\CSV output".


INFORMATION AND CONTACT
-----------------------
This Methodology file was written in August 2023 by David Katzin, Greenhouse Horticulture and Flower Bulbs, Wageningen University and Research. david.katzin@wur.nl

This file is part of the dataset "Data from: ‘GreenLight - An open source model for greenhouses with supplemental lighting: Evaluation of heat requirements under LED and HPS lamps (2020)’"
DOI: 10.4121/78968e1b-eaea-4f37-89f9-2b98ba3ed865
URL: https://doi.org/10.4121/78968e1b-eaea-4f37-89f9-2b98ba3ed865