Drones and remote sensing technology, such as multispectral and thermal cameras, have the potential to detect early symptoms of stress in turf grass. A pilot project was conducted to determine if or how this technology could help sports turf maintenance. This article focuses on the early detection of the turf grass disease Grey Leaf Spot, which is caused by the harmful fungus Pyricularia spp
Figure 2: Field 11 in North-South orientation with area in question (white outline)
The St. Jakob sports facility in Basel was flown over by a drone equipped with a multispectral camera once a month. The remote sensing images were evaluated, compared with observations on the ground and conclusions were drawn to inform ongoing maintenance.
Field number 11 of the junior campus of Basel Football Club was severely affected in August due to the rapid spread of the fungus, and images taken with a multi-spectral camera, in combination with the NDVI index, allowed early detection before the damage became visible to the human eye.
Sports fields and golf courses are often criticised for their use of pesticides, fertilisers and water consumption. If turf problems are detected at an early stage, it should be possible to strengthen the plants through mechanical procedures and biological inputs and to avoid excessive addition of chemicals, fertilisers and water. Remote sensing techniques could be beneficial in this early detection. To this end, an initial pilot project, lasting over a year, was launched at the end of 2018 in collaboration with the Sports Department of the City of Basel and FC Basel 1893.
The work was undertaken between February 2019 and January 2020. The aim of this pilot project was to investigate the potential applications of remote sensing technology in sports turf maintenance, using practical examples to find out how effective it is in the early detection of plant stress - with the ultimate aim of reducing the use of pesticides and irrigation.
Figure 1: Flight perimeter at the St. Jakob sports complex in Basel with Field 11
Materials and Methods
Locations and test set-up
In the summer of 2011, the field 11 (main FCB match field - Figure 1) at the St. Jakob sports facility in Basel was completely rebuilt as a Lavaterr pitch, including new drainage pipes, in-ground heating, automatic irrigation and seeded with a sports turf mixture consisting of Lolium perenne and Poa pratensis. After eight years, the pitch also contained a high percentage of Poa annua. In autumn 2018, the entire middle third of the pitch was repaired and sodded with new turf with a high proportion of Lolium perenne (see dark green area in Figure 2). To begin with, a test flight was undertaken over the entire St. Jakob sports facility in order to establish a "zero measurement" and 16 georeferenced control points (GCP), measured in Basel and stored in the project database in order to locate, overlay and compare the flown results precisely.
Following the test flight, the official data collection flights were flown once a month with the drone (DJI Matrice 210), equipped with both multispectral camera sensors (MicaSense RedEdge-M) and an RGB camera (DJI Zenmuse X5s). All recordings were made automatically, using pre-programmed flight paths and the following parameters: flight altitude of 80m, overlap of 78% / 78%, 20 rows (north-south), total distance covered ~14km, flight time ~38 minutes (2 flights + battery change), approx. 8'800 single images per flight (5 channels with 1700 images each). Flights were always done in the middle of the day between 12:00 to 13:00 and, to compensate for any changes in the light quality, a calibration measurement was carried out before and after the flight using a reference disc from MicaSense. This made it possible to adjust the acquired images to the specified values during processing and to obtain a comparable result.
Multispectral camera and processing of the acquired data
The images were captured by the five spectral-specific designed and calibrated lenses and stored as jpg files. Multispectral image data is composed of several spectral channels that capture reflected electromagnetic radiation in green (497-530nm), red (620-780nm) and near infrared light (780-1'400nm), however, before processing, the image files initially appear in pale grey scales. The representation in the known colour gradients of the NDVI (Normalized Difference Vegetation Index) results from the algorithm-based calculation of the individual spectral ranges and the programmed colour presets. The final NDVI images illustrate the reflection of the green, red and near infrared light reflected off the turf surface, with greener colours indicating healthier plants with, for example, a greater percentage of chlorophyll in the plant leaves, and red indicating stressed or sick plants with a weaker vitality. The photogrammetric evaluation was then done with Pix4Dmapper and from there the software Pix4Dfields allowed the superimposition and comparison of recordings and indices. After processing, the data was uploaded to Pix4Dcloud to make it available to all participants via the browser.
Left: Image captured on 02.12.2019, RGB Orthophoto Right: Image captured on 02.12.2019, NDVI-Index
Meteorological data and weather conditions during the recordings
At the same time, weather data was obtained from the Federal Office for Meteorology and Climatology (MeteoSwiss) via the BAS Basel/Binningen weather station at 316 metres above sea level, and the following measurement parameters were used for the evaluation: precipitation in mm, temperature 2m above ground in C°, temperature 5cm above ground in C°, sunshine duration in hours, relative humidity 2m above ground in % and wind speed in km/h. The comparison of the weather data with the maintenance plan of field 11 was an important basis for the analysis of the plant indices.
Maintenance, irrigation and occupancy
Also, all maintenance measures, with the exception of irrigation, were reported daily and were thus available for further processing and evaluation (see Figure 3). To date, the greenkeeping team has not yet collected any data on irrigation, so no evaluable figures on irrigation days and quantities were available. The maintenance data received from the greenkeeping team included mowing intervals and cutting direction, aeration and sand topdressing, the various fertilizer applications, the use of pesticides and wetting agents, over-seeding, etc.
Evaluation of NDVI data
In the analysis of the NDVI data during the period July 2019 to November 2019, the change in plant vitality and growth can be clearly identified by the discoloration from the green to the red colour range. As determined by the software and user preferences, the greener the colouring, the healthier the sports turf is, the redder the more the turf sward suffers from stress. The sequence of pictures below illustrates (through this evolution of colour) how dramatically the grass stand changed over almost 4 months under the influence of the fungus Pyricularia spp.; the highlight of the spread being at the beginning of September.
The NDVI measurements indicate that the grass plants were under significant stress between the period of July to September and this stress was not uniform across the entire field. In comparison with the climate data during this same period, it could be surmised that the high temperatures during the summer months, between the end of June and early September, were a major stress factor for the grasses, specifically in the older turfed area. The heat tolerance of the different grass species would appear to be different and, if we look at the areas on the pitch which were most strongly stressed, the high temperatures seemed to put more pressure on the older areas which contained a high percentage of Poa annua. On August 5, 2019, the multispectral data on this older turf illustrates the severe decline in vitality of the overall grass sward containing a high percentage of Poa annua.
Figure 3: Diagram showing weather data and maintenance practices on Field 11
On the other hand, the newly sodded area in the centre of the pitch shows a significantly better vitality up until August 5, despite the stress signs due to the pressure of heavy use. Only the northern area towards the side line shows an anomaly, due to unknown factors, but which may be the first signs of a general change in the vitality of the Lolium perenne cultivars (signs which are visible in the NDVI index images).
Looking back, temperatures had risen sharply from June 22 and excessive artificial irrigation was necessary because of the lack of rainfall, creating conditions conducive to the occurrence of Grey Leaf Spot (L.B. McCarty, 2005).
On August 6, these stressed areas were analysed on site and identified by Bernhard Schenk (Division Manager at UFA-Samen PROFI GRÜN) as the turf disease Grey Leaf Spot.
By September, the area covered with sodded turf had been almost completely destroyed by the Grey Leaf Spot. The clear demarcation from the older area with a significantly lower proportion of Lolium perenne is clearly visible, illustrating that Lolium is affected by Grey Leaf Spot, whereas Poa species seem to be less affected. Results which are corroborated by L.B. McCarty (2005).
Evaluation of weather data and maintenance measures
The weather data from the weather station BAS Basel/Binningen shows the last precipitation on June 20 with 26mm. After that time, the weather turned hotter and dryer with many hours of sunshine and only sporadic precipitation was measured between the end of June and the beginning of August. Due to the low rainfall, the pitches had to be irrigated daily in the early morning hours. The measured wind movements can be defined for the entire period of the measurement as a light breeze (2-9km/h) to a gentle breeze (10-19km/h), providing little air for surfaces to dry out following irrigation. During the same period, the greenkeeper cut less often and, on June 24 and July 27, fungicides were applied due to a summer fusarium infestation. The site was over-seeded for the first time on June 28. The hot, dry period continued into July and with total rainfall of only 80 mm the need for irrigation remained high. Relative humidity remained high, averaging 61%. On July 5, the entire site was over-seeded again and on July 8 it was aerated and then sanded. On July 10, the site had to be over-seeded a third time and on July 11 it was fertilised.
Overview of the NDVI measurements on Field 11
By early July 2019, conditions in Basel were theoretically ideal for the Grey Leaf Spot. It was hot and humid and the turf was intensively irrigated and kept moist for a long time. The pitch had been freshly fertilised and there was no strong air movement on and around the pitch to dry it out following morning irrigation. Although the Grey Leaf Spot disease probably started at this time, the NDVI image from July 4 doesn't necessarily reveal any particular indication that would point to a disease outbreak. The images indicated that much of the pitch was stressed, likely due to the hot dry conditions and wear and tear from daily practice. It was therefore difficult to differentiate between stress due to wear and tear or weather conditions and the beginnings of a disease outbreak. However, the August 5 image (below) does seem to show the beginning of some stress on the north-west corner of the central portion and, looking back, this is likely the first signs of an infestation on the Lolium perenne cultivars of the newly sodded area, which was confirmed by Bernhard Schenk of UFA-Samen on August 6. The September 4 NDVI image clearly shows that the outbreak had almost exclusively infected the new Lolium perenne sods and hardly touched the older surface high in Poa species.
In fact, on July 27 the greenkeeper had sprayed for Fusarium which had affected parts of the older Poa surfaces, not yet knowing that Grey Leaf Spot was also present. If the greenkeeper had received this additional knowledge earlier, he might have been able to react in time with some practical measures which could have saved or at least mitigated the spread of the disease over the central sodded portion of the pitch. However, due to the widely spaced flights, one month apart, the remote sensing images were only partially useful and it became clear that one flight per month in the main growth period, and in the months with a high incidence of disease, is far from sufficient to respond quickly and effectively to turf problems.
Nevertheless, the analysis of the monthly recorded images revealed clearly visible damage caused by stress and, overall, a clear visualisation of the vitality of the grass sward. The maintenance measures of the greenkeeping team and the weather influences could also be made visible and comprehensible.
The case study presented here, with a possible early detection of the turf disease Grey Leaf Spot, illustrates how valuable it is when a greenkeeping team becomes aware of anomalies on sports turf in good time and can react accordingly, especially with a turf disease like Grey Leaf Spot which spreads rapidly and can only be chemically suppressed at an early stage. Overall, the comparison of the NDVI images with the on-site inspections was very revealing. Based on our observations, the camera images were clearly able to indicate a weakened vegetation earlier than the human eye was able to see changes on the ground. To conclude, vegetation indices and their evaluation, in combination with an analysis of local weather data and applied maintenance measures, could prove to be a valuable tool for future-oriented and resource-saving sports turf maintenance.
Field 11 on August 5, 2019 (NDVI-index and RGB image)
Roland Berger, David Bily, Christian Desgranges and Erich
Steiner - Steiner & Partner Landschaftsarchitektur GmbH
Reto Weiss - pixmap gmbh
ALTHOFF, M. (2019): Drohnen im Golfclub-Alltag. Greenkeepers Journal, 2/2019.
CAMPBELL, J.H., HENDERSON, J.J., MORRIS, T.F., GUILLARD, K., INGUAGIATO, J.C., RACKLIFFE, S.L. AND WALLACE, V.H. (2015): The effect of nitrogen source and total nitrogen on color and quality of Kentucky Bluegrass. - University of Connecticut, College of Agriculture, Health and Natural Resources, 2015 Annual Turfgrass Research Report: p. 64-66.
CATUREGLI, L., GROSSI, N., SALTARI, M., GAETANI, M., MAGNI, S., NIKOLOPOULOU, A.E., BONARI, E. AND VOLTERRANI, M. (2015): Spectral Reflectance of Tall Fescue (Festuca Arundinacea Schreb.) Under Different Irrigation and Nitrogen Conditions. Agriculture and Agricultural Science Procedia 4, p. 59-67.
GILL, J. (2018): Developing a Drone Strategy for Use in Sports Turf Management (https://sportsturfonline.com/2018/01/16/developing-a-drone-strategy-for-use-in-sports-turf-management/).
HALL, R. (2016): How Drones Could Transform the Landscape Industry (https://www.turfmagazine.com/technology/drones-in-landscaping/).
ISHIMWE, R., ABUTALEB, K. AND AHMED, F. (2014): Applications of thermal imaging in Agriculture - A Review. Advances in Remote Sensing, 3, 128-140.
LICHT, B. (2019): Drohneneinsatz im Greenkeeping. Greenkeepers Journal, 2/2019.
MCAULIFFE, K. AND ROCHE, M. (2011): Developing new innovative technology for benchmarking turf grass performance. Project Number: TU09039. - Horticulture Australia Ltd (Sydney, Australia).
MCPHERSON, D. (2016): Drones, data and turf maintenance (https://www.golfdom.com/drones-data-and-turf-maintenance/).
PEACOCK, C.H. (2008): Monitoring for pest problems, a new UAV perspective (www.turffiles.ncsu.edu).
WACHENDORF, M., FRICK, T. AND MÖCKEL, T. (2017): Remote sensing as a tool to assess botanical composition, structure, quantity and quality of temperate grasslands. Grass Forage Sci. 2018, 73:1-14.