Spatially scaled and customised daily light integral maps for horticulture lighting design

ABSTRACT

In order to produce high-quality plant materials, it is necessary to consider a series of biotic and abiotic inputs. In this study, we focus on horticultural light climate mapping in Europe using earth and weather observation and information technologies. It is widely accepted that optimised light programmes and spectral recipes with dedicated wavelengths can most efficiently support the photoreceptors and the controlled plant production. By knowing the DLI (Daily Light Integral) for a particular area, growers can optimise their crop management strategies, such as selecting the most appropriate crops for the light levels, determining optimal planting times, and selecting the best location for their crops. Generating a sufficiently resolved DLI map usually demands both a spatial and spectral downscaling process. In our present research we thus focus on (i) the development of a semi-automatic DLI mapping workflow and (ii) a first exemplary visualisation of an all-season DLI map for a European country, which can easily be adapted to any other country based on the suggested approach. A special focus was put on the development of precise DLI values at a European scale, especially experiencing with 1 and 2 mol·m⁻²·d⁻¹ DLI value increments. Additional purposes of DLI mapping such as the adaption to climate changes and the efficient use of energy were also addressed as the present energy crisis documents the need to adapt future horticulture engineering systems both for indoor and outdoor production. Customised DLI maps are also useful secondary information sources for solar energy mapping, especially for renewable energy sources.

KEYWORD:

• DLI charting
• Europe
• map scaling
• plant growth LEDs

1. Introduction

A rising interest in horticulture data mapping is experienced since spatial data acquisition, earth observation techniques and networked smart sensors are available. Horticulture professionals could benefit from customised production information at a given location or a well-defined region. Currently, generating a DLI map is tied to a spatial downscaling of traditional and spaceborne data, which can be customised to continental, regional or local (farm level or greenhouse) needs

According to historical scientific records, DLI is gaining an increasing interest in horticultural research and commercial developments. Due to Web of Science data collection, there are more than 550 scientific papers exactly related to the term “daily light integral”. In Figure 1 a trend and a clear potential for horticultural lighting research and industrial development can be seen. Since 2014, the number of contributions has been continuously in double digits, and in 2022 there are exactly 51 contributions, with a tendency for further increases.

Figure 1. Web of science-based trend concerning „daily light integral” papers between 1983 and 2022 (date: 09.01.2023).

As the DLI is a key factor in the production of cultivated plants (Faust & Logan, 2018), its impact on horticultural crops needs to be investigated in a number of aspects, including photosynthesis, morphology, yield, nutritional quality (Dou et al., 2018), the growth of vegetative parts of seedlings and cuttings (Hernández & Kubota, 2014; Yan et al., 2019), the induction of flowering (Christiaens et al., 2014; Warner & Erwin, 2003; Whitman et al., 2022), or the determination of optimal DLI values for a given plant species (Higashide & Heuvelink, 2009). In plant experiments, the DLI represents an important treatment for experimental setups (Shao et al., 2022).

DLI has an exact scientific definition. It is an accumulation or integration of quantum flux measurements per second over 1 day (24 hours), traditionally in the photosynthetically active radiation (PAR) spectrum normally located in the 400 to 700 nm wavelength range (see Figure 2). Its widely used unit is mol·m⁻²·d⁻¹, which is provided by the photosynthetic photon flux density (μmol·m⁻²·s⁻¹) × photoperiod (h·d⁻¹) × 3600 (s·h⁻¹) × 10⁻⁶ (Korczynski et al., 2002; Yan et al., 2019). Its definition is well described, while the methods to retrieve it often differ. Korczynski et al. (2002) and Thimijan and Heins (1983) worked with the assumption that the photosynthetically active radiation (PAR) from 400 to 700 nm takes 42.9% of the total solar radiation between 300 and 3000 nm. This reduction leads to PAR solar energy in W·m⁻², which value is multiplied by 4.57 to convert PAR solar energy into PAR photon flux in μmol·m⁻²·s⁻¹. Based on the latest studies of Blonquist and Bugbee (2017) and Faust and Logan (2018) corrected the results of Korczynski et al. (2002) and assumed that 45% of the total solar radiation is contained in the PAR and multiplied the solar PAR W·m⁻² values by 4.484 μmol·m⁻²·s⁻¹. The same approaches are adapted by instrument manufacturers to provide DLI measurements (LI-COR LI 1500, 2023; Apogee DLI 400, 2023; Quantum PAR/DLI, 2023) for greenhouse growers. In our spatially distributed calculations of DLI and the seasonal DLI mappings (see Figure 2) we followed the principles of Faust and Logan (2018).

Figure 2. Seasonally averaged DLI values of different cities in the northern hemisphere.

It is beyond the scope of our study to conduct a deep market research on horticulture lighting industry, but the growing scientific interest is obviously linked to the present and future LED farming potential globally arising from the vivid greenhouse, vertical and indoor farming and lighting developments. In this paper we thus focus on (i) the development of a semi-automatic DLI mapping workflow and (ii) the first visualisation example of an all-year DLI map for a European country (Hungary).

2. Materials and methods

We performed our tests and calculations for a Central European country, Hungary, chosen as region of interest (ROI) in our study. Nevertheless, the approach that we follow is geographically independent and open, so that it could get applied to any other country or region.

In the present ROI more than half of the studied area are plains below 200 m above sea level (a.s.l.), and the area above 400 m a.s.l. is less than 2% (HMS, 2022). The horticulture production areas of Hungary are typically situated in the Great Hungarian Plain, which covers approximately 52,000 km² of Hungary, about 56% of its total area. The highest point of this plain is at 183 m a.s.l., the lowest point is the Tisza River.

DLI data calculations for the ROI were based on the DLI engine of SunTracker Technologies Ltd, Canada. Our raw data access and query were processed in communication with the company developers. The calculation methods applied were based on the work of Faust and Logan (2018) that uses worldwide weather station and satellite data. In case of missing satellite data at a particular location these data were replaced by interpolated World Meteorological Organization (WMO) ground weather station data. For the ROI, a total of six WMO weather stations were available, two domestic ones (from west-to-east: Szombathely és Debrecen) and four neighbouring ones close to the borders of Austria, Slovakia and Romania. Calculations were continually qualified and unreliable data were extra marked if there were no weather stations located within 100 km of the selected location and there were missing satellite data (GOES NOAA) for that location. In such cases the predicted DLI values may not be accurate and the user is warned to use the data. In our case no such warning was experienced.

For a reasonable spatial sampling strategy, we applied a spatial data grid, which is globally used for digital elevation models (DEM). We referred to the grid of the 1-arc second global DEM of the Shuttle Radar Topography Mission (SRTM), which realizes an approximately 30 m spatial resolution on the ground (projection: Geographic GCS, Horizontal Datum: WGS84, Vertical Datum: EGM96). The 30 m grid resolution was used as spatial sampling template and its coordinates as DLI spatial sampling points in the ROI area (Rodriguez et al., 2006).

The basic processing task was to create a script to retrieve data from a remote server, based on given DLI parameters, and to automate the process due to the bulk of data. The script retrieves DLI values for a given geographical area based on an SRTM resampled grid positions. The geographic latitude and longitude values are sent by the device running the script (text processor readable format, CSV) in the form of a GET parameter in the URL calling the remote server. The server serving the dli.suntrackertech.com URL performs the necessary computations and delivers the result to the querying devices that initiated the request. The Application Programming Interface (API) listens for requests on port 8443. For example, the URL to retrieve DLI values for a given geographic point is “https://dli.suntrackertech.com:8443/DLI/api/get_DLI/48.43,-123.36‘, where ’48.43‘ is the latitude and ’−123.36” is the longitude. The remote server (server hosted by SunTracker Technologies Ltd.) returns the response in a JavaScript Object Notation (JSON) format, which contains the DLI values for that geographic area on a monthly basis. The script stores these values, appending them to the source CSV file, which is saved as a new (result) file. For this task we used a Linux operating system (Linux Mint 20.3, Ubuntu Server 20.04.4 LTS on Oracle VirtualBox virtual machine and Windows Subsystem for Linux) and PHP language.

Several considerations had to be taken into account to implement a processing flow to avoid an overloading of the server performance and the computation capacity (i), to capture the inevitable missing/incorrect responses (ii) for later reprocessing and to automate a script recognising stop and initiate restart continuously (iii) to avoid data loss.

For point (i), it is important to mention that the process was preceded by a consultation with SunTracker Technologies Ltd., which gave its approval for the operation. The requests were made from multiple computers on different networks (up to three at a time), without delay, running as a single process on a single computer at a time.

The condition described in point (ii) was fulfiled by storing the erroneous (no response received, wrong/incorrect response received) responses as a special, clearly identifiable string (“xxxx”). By running an additional script (repeat.php) on the result CSV file it was possible in most cases to filter out these lines, retrieve the data again and to save the actually correct values in one step. A final check was performed to determine if there were any “xxxx” strings left in the result file – if necessary, these could be corrected by running the “repeat.php” script again.

For the implementation of point (iii) (resumption capability), checking the last line was crucial, since resumption is only correct if there is no discernible break in the script (i.e. no incomplete or erroneous lines between sections). Thus, the last line must either be complete and error-free, or it must be completely deleted to obtain a result file with a complete and error-free last line. The script decides on the deletion by checking whether the formal requirements for correct lines are met, which is solved using a regular expression. Regardless of whether a deletion was made or not, the process continues with the same step: The script looks for the line number of the last still retrieved line with the correct answer and continues the query from there (Figure 3).

Figure 3. Flow-chart for DLI data collection.

3. Results

DLI is the accumulation of per second quantum flux measurements over one day, so that it seems to be advisable to represent the daily flow over a longer period of time to characterise conditions of supply. We therefore present, similar to Faust and Logan (2018), the accumulated monthly quantum flux values in Julian calendar style. The studied country located in the mid-latitudes and extending over 6° longitude (16°09’E, 22°51’ E), shows both distinct regional differences and significant seasonal photon flux changes (Figures 4 and 5). The accuracy of the results depends in general on the number of input data; DLI varies depending on factors such as location, topographic diversity, season, or weather conditions. For instance, more than 2000 sunny hours a year are common in the southern and south-eastern parts of the ROI. For visualisation and interpretation purposes Figures 4 and 5 differ in data scaling resolution (1 and 2 mol·m⁻²·d⁻¹, respectively). Analysing the seasonal fluctuations of the two differently scaled DLI maps shared characteristics can be observed. The highest month-to-month variation of DLI values can be seen in the summer seasons. Summer DLI values range from a minimum value of 33 mol·m⁻²·d⁻¹ up to 47 mol·m⁻²·d⁻¹, while the wintertime DLI starts at 4 mol·m⁻²·d⁻¹ and reaches 14 mol·m⁻²·d⁻¹, respectively. Using a 1 mol·m⁻²·d⁻¹ scale more spatially reasonable differences can be shown, especially for the spring and summer time of the given region.

Figure 4. Monthly DLI maps of region of interests (ROI) in Hungary with 1 mol·m⁻²·d⁻¹ DLI value increments. County borders are delineated within the country.

Figure 5. Monthly DLI maps of region of interests (ROI) in Hungary with 2 mol·m⁻²·d⁻¹ DLI value increments. County borders are delineated within the country.

In their differentiation, the results presented in Figures 4 and 5 show that country-level DLI maps for Europe require some special considerations. Unlike a large-scale continental DLI map (Faust & Logan, 2018), it is challenging to highlight and present differences at an appropriate scale arises for a single country. Figure 4 documents a marked seasonality, with lower values in the winter months (4–5 mol·m⁻²·d⁻¹) and higher values in the summer months 45–50 mol·m⁻²·d⁻¹).

Therefore, as an alternative to Figure 4 and following the widely used DLI map for the United States (Faust & Logan, 2018), scaling in 5 mol·m⁻²·d⁻¹ increments was used to represent DLI conditions in the ROI (see Figure 6). However, small-scale DLI variations were no longer apparent, whereas large-scale differences were still clearly evident. For the representation of local to regional differences, however, a scaling with 1 mol·m⁻²·d⁻¹ DLI increments seems rather recommendable to serve professional needs with a focus on horticulture.

Figure 6. Monthly DLI maps of region of interests (ROI) in Hungary with 5 mol·m⁻²·d⁻¹ DLI value increments. County borders are delineated within the country. A Faust and Logan (2018) style scaling and labelling for comparison purposes.

4. Discussion

DLI in Hungary can range, according to our calculations, from about 4–5 mol·m⁻²·d⁻¹ in winter up to 46–47 mol·m⁻²·d⁻¹ in summer, depending on specific local conditions and prevailing weather conditions. For mid-latitude countries (approximately 23° to 66° in the Northern and Southern Hemispheres), additional LED lighting could be developed specifically for use in the early spring season for application in sunlit indoor planting, accelerated growth, or improved growing and breeding conditions. In the ROI presented, early spring (e.g. March in Hungary) is a critical production period with marked lighting deficits, while light surpluses in the summer season (e.g. June–August in Hungary) can lead to shading problems, for example.

Common features of LED lights for crop production could include an adjustable light spectrum, dimming capability and energy efficiency. Many early-growing vegetables grown in greenhouses would benefit from an optimised lighting management to mature quickly and to be harvested within a relatively short period of time after planting. In this regard, one should be aware that DLI is only one factor determining the overall success of vegetable production. Other factors such as temperature, humidity and nutrient availability also play a crucial role in the growth and development of plants, so that an integrated management seems reasonable. It is worth noting again that the actual indoor and outdoor DLI will depend on a number of local factors, such as cloud cover, atmospheric conditions, greenhouse construction and glazing materials, and the presence of any shade or obstructions (Kelly et al., 2020; Palliwal et al., 2021; Seginer et al., 2006).

To determine the additional lighting needs for greenhouse conditions, case-specific conversion factors are needed. The variation in light levels can be significant, influenced by shading or variations in reflectivity due to the nearby terrain.

Early vegetable production involves growing vegetables in greenhouses or other controlled environments in the off-season or early season, using artificial light to compensate for the lack of sunlight. DLI is an important parameter in early vegetable production because it represents the total amount of light a plant receives in a day, which is critical for plant growth and development. In the early or late season, natural sunlight may not be sufficient to achieve the DLI required for optimal plant growth. For this reason, growers often use artificial lighting systems to achieve the required DLI value. The optimal DLI for early vegetable production depends on the type of crop and growth stage but is usually between 10 and 25 mol·m⁻²·d⁻¹. Baumbauer et al. (2019) found that increasing DLI had a positive effect on lettuce fresh weight; it increased from 1.27 g/plant to 4.33 g/plant. Dry weights of all species increased linearly with increasing DLI, with lettuce increasing 203%, kale increasing 47%, and spinach increasing 42% when DLI increased from 8 to 14 mol·m⁻²·d⁻¹. Based on the recommendations of Runkle (2019), some DLI target values are summarised in Figure 7. It should be noted that target DLI values can vary significantly depending on many factors such as shade tolerance and environmental factors (including temperature and CO₂).

Figure 7. Optimum DLI and PPFD ranges with plant usability (based on and modified after Runkle, 2019).

DLI maps generated in this study document the effective natural light conditions. To achieve optimal DLI values for early vegetable forcing, growers can use high-pressure sodium (HPS) lamps or light-emitting diodes (LEDs) as artificial lighting systems. They can adjust the spectrum, intensity and duration of the light with the aim to maintain a consistent DLI throughout the growing season and ensure the right quantity and quality of light for optimal growth and development conditions.

5. Conclusion

Climate change, growing urbanisation, growing food demand, extreme seasons, food security and economy pressure promote the expansion and consideration of LED-based production systems. In order to produce high-quality plant materials, optimised light programmes are needed. Using the dedicated wavelengths, the photoreceptors can maximally utilise the available resources. Due to the present energy crisis and future demands, energy efficiency is a key term in future horticulture engineering systems both for indoor and outdoor applications. In this development DLI maps will help farmers better understand their light patterns and needs (Balázs et al., 2022; Sipos et al., 2020). Listed below are several DLI map considerations outlined in the following short sections.

5.1. Spectral light measurements

In horticulture plant production light is one of the driving abiotic limiting factors. In the past (before the horticulture LED era) more attention was paid to light quantity than light quality, because spectral measurements and spectrally tuned light sources were less available. To ensure precise light control of plant metabolic reactions in controlled environments, it becomes essential to deepen our understanding of spectroscopy. This is because upcoming DLI measurements must rely on more comprehensive insights into the spectral properties of light. The specific light requirements of plants, LED light recipes, and incoming light distributions are all quantified in terms of photons, highlighting the need for a photon-based approach to optimise plant growth. Traditional PAR approaches will likely be extended (e.g. utilizing ePAR or UV-B photoreceptors), since LED manufacturers are able to go beyond and below the visible light by extending the spectral range recommended by available research results in plant biology (Kami et al., 2010; Zhen et al., 2021). Therefore, the PAR-based photon flux values will likely be reviewed and tuned accordingly.

5.2. DLI map as “light compass”

The ability to calculate spatial DLI values opens up new possibilities. Horticulture professionals can estimate how much supplemental lighting is needed during the year to optimise potential plant growth and economic efficiency in a greenhouse. Farmers can also better manage the shading needs for their crops. Even outside the greenhouse the DLI map is a kind of “light compass” to better determine whether field crops will receive sufficient photon flux during the phenological stages to reach optimum maturity.

5.3. Individualisation in horticultural production

Modern horticultural production will continuously be individualised. Customised treatment strategy monitors relevant individual organs and changes in real-time. The higher temporal and spatial resolution in DLI data acquisition will be merged and the farmers will benefit from this synergy for field crops, traditional greenhouses or controlled-environment horticulture.

5.4. DLI map linked solar energy chart

Horticulture is very resource consuming, climate change and the evolving energy crisis became relevant production drivers, which negative impacts should be minimised and reduced. A spatially detailed DLI map can be used as a solar energy chart to better understand and plan sun energy availability. For this case some technical modifications and different geometrical approaches are needed, because solar panels are typically not horizontally situated and intend to increase incident light intensity without plant physiological considerations. These modifications are available and used by most solar energy designers and can be linked to DLI maps.

Acknowledgments

Special thanks go to Ian Ashdown, senior scientist at SunTracker Technologies Ltd, Canada for answering our research questions and supporting our initiative with ideas. A. J. and Zs. V. were supported by project no. TKP2021-NVA-29, which has been implemented with the support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, financed under the TKP2021-NVA funding scheme.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the National Research, Development and Innovation Fund [TKP2021-NVA-29].