THESIS OF DOCTORAL (PhD) DISSERTATION ISTVÁN BORSICZKY MOSONMAGYARÓVÁR 2018
SZÉCHENYI ISTVÁN UNIVERSITY FACULTY OF AGRICULTURAL AND FOOD SCIENCES MOSONMAGYARÓVÁR ANTAL WITTMANN MULTIDISCIPLINARY DOCTORAL SCHOOL OF PLANT, ANIMAL AND FOOD SCIENCES HEAD OF DOCTORAL SCHOOL: PROF. DR. VINCE ÖRDÖG, DSc GOTTLIEB HABERLANDT PLANT SCIENCES DOCTORAL PROGRAM PROGRAM LEADER: PROF. DR. VINCE ÖRDÖG, DSc SUPERVISOR: PROF. DR. REISINGER PÉTER, CSc EFFECTS OF SITE-SPECIFIC WEED CONTROL BASED ON SENSOR TECHNOLOGY ON CHANGES OF THE WEED FLORA OF ARABLE LANDS AUTHOR: ISTVÁN BORSICZKY MOSONMAGYARÓVÁR 2018
1. INTRODUCTION AND OBJECTIVES The heterogeneity in agricultural areas are a challenge and opportunity for farmers at the same time. The challenge is the mapping of heterogeneity, the opportunity is the possibility of using site-specific technology to provide variable doses within the field. The possibilities for the use of current precision protection technologies within different groups of pests were compiled by OERKE ET AL., (2010) (Table 1). Table 1. Possibility of using precision technology within different groups of pests OERKE ET AL., (2010). Trait Weeds Nematodes Insects Pathogens Size of organism (mm) 1-1000 0,1-1 0,1-00 0,0001-1 Cycles per season 1 1-5 1-8 (?) 1-9 (?) Mobility Very low Low Low to high High Field heterogenity Detection XX(X) XX(X) X(X) X(-) Individuals (XX) Disease symptoms Individuals, sympt. (X) Disease sympt. (X) Identification XX -?? Quantification XX (X) (X) (X) Prognosis/DSS X(X) X(X) (X) Data management Off/On-line Off-line Application technique XX(X) (X) (X)? XX advanced stage; X first steps/moderate knowledge;?not known/not feasible; - knowledge low Currently in the most advanced stage of development is the protection against weeds. The size of the organisms facilitates detection and 1
recognition, while their very low mobility allows the use of site-specific off-line control technology against them. GNSS positioning instruments and sensor technology, as well as robotics, have provided new opportunities for our research. The availability and affordability of autonomous vehicles (drones) led us to examine their practical usability in the site-specific weed control technology. Fertilizer application based on remote sensing data and sensor use is increasingly spreading through the expansion of the professional advisory sector. However, the sensors are measuring the vegetation indices of the crops together with the weeds and we have examined their possible distorting effects. High value-added products produced in organic farming could be a competitive advantage for domestic farmers in the long term. The availability of tractors and implements driven by sensors and positioning devices can open new opportunities for the practice. 2
The main objectives of the dissertation: This dissertation is about an important topic of digital agriculture, the possibilities and directions for developing site-specific and precision weed control methods. 1. Development of the Balázs - Ujvárosi weed survey method based on autonomous vehicle 2. Examination of the distorting effect of weed vegetation on the NDVI vegetation index for cereal grain 3. Examination of precision agriculture s implements and tools in the practice of organic weed control Reference (OERKE EC, GERHARDS R, MENZ G, SIKORA RA (EDS.), 2010: Precision Crop Protection- the Challenge and Use of Heterogenity, Springer ISBN 978-90-481-9276-9) 3
2. MATERIALS AND METHODS 2.1. Use of autonomous vehicles in the site-specific weed control practice Use of autonomous aircraft We have started our investigations with an autonomous aircraft. As an experimental device we have used a DJI S900 professional UAV hexacopter in autonomous mode. For capturing the images, we used a Canon PowerShot SX260 HS camera with a 12-megapixel resolution and optical image stabilization. The aircraft progressed at a height of 5, 10 and 20 meters, according to the programmed route and stopped to capture images at the sampling plots. The construction process of the autonomous weed-mapping robot For the safe detection of germinating and overlapping weeds, we first aimed to improve the resolution and quality of the images. As a synthesis of our weed mapping experiences we have started to build the autonomous weed-mapping robot in 2016. The prepared vehicle operates on the ground in autonomous mode according to the GPS coordinates of the weed-mapping plan and creates submillimetre-resolution images of the plants at the programmed sampling points. In the experimental area we carried out two types of weed mappings, first in accordance with the improved Balázs - Ujvárosi method, at 0.5 hectares frequency, 21 sampling points, 2 2 meters sampling area. Secondly, the autonomous vehicle carried out the weed-mapping in accordance with the programmed algorithms. The 326 recordings were evaluated under office conditions. 4
2.2. Relationship between NDVI indexes and weed coverage in winter wheat The relationship of weed-mapping and vegetation indices In the spring of 2013, during the survey of precision weed control observation plots in winter wheat, a Trimble GreenSeeker vegetation sensor was used for crop monitoring. By analysing the data with statistical methods, we have concluded a correlation between the canopy coverage and the NDVI vegetation index. Measurement of the vegetation indexes of the plots before and after hand weeding In the spring of 2014, we have randomly selected 20, 1 1-meter sampling plots in winter wheat. With a Trimble GreenSeeker Handheld instrument, we have measured the NDVI indexes of the sampling plots. After hand-weeding we have reexecuted the instrumental measurements. We have included our data in tables, and then we have assigned nitrogen top dressing fertilization doses for the measured values based on the NDVI values. 2.3. Weed control experiment in organic grown sunflowers For our experiment we have used a precision guided inter-row cultivator equipped with finger weeder, a tractor with RTK steering and a precision planter. 5
On the experimental field we have selected 4 10 test plots of 6 6 meters and we have performed three types of treatments in random distribution. We have left untreated control areas and we have treated with the inter-row cultivator, and as the third kind of treatment the inter-row cultivator was equipped with rotary finger weeders. After the treatments we carried out weed surveys and the results were processed and evaluated under laboratory conditions. The area was subsequently treated again and height measurements on randomly selected sample plots were carried out. The plants from the experimental plots were harvested manually, the anthocarps were measured by the mass and diameter, and trashed using a plot combine harvester. We have processed and evaluated the results. 6
3. RESULTS 3.1. Use of autonomous vehicles in the site-specific weed control practice The obstacle of large scale spreading of the off -line site-specific weed control is the absence of an internationally standardised weed-surveying method. The widely used Balázs - Ujvárosi method in Hungary has been adapted to a weed-mapping field robot. The robot can eliminate the exhausting field work. The robot is environmentally friendly, electrically driven and does not cause damage to the vegetation. It is equipped with modern GNSS positioning tools and operates at high level of automation even at extreme weather conditions and the collected photos have high resolution. The photos are made at the wavelength of the visible light, therefore there is no need for conversion, the images can be evaluated by an expert in office conditions. The equipment is patented: (G02B 27/00 A01M 7/00 A01M 21/00). The autonomous weed-mapper can be an element of digital agriculture. The robot performs a pre-programmed sequence on the planned route in the field, and the density of the samples (photos) are high. The recordings are digitally transmitted to the cloud-based database and then from there to the high-resolution screen of the evaluating professional herbologist or agronomist. After image processing, the weed texture of the field is recorded, based on which the site-specific weed control technology can be executed successfully. The specificity of our method is that the weed detection process is not fully automated, the recognition of weed species is carried 7
out by a herbologist and thereby the software recognition inaccuracy is eliminated. After the identification of the species and weed coverage, the data will be restored to a digital environment, where the site-specific weed-control technology is established based on the algorithms developed by us. The additional advantage of our off-line method is that image processing takes a short period of time, and therefore takes only 1-2 days between the implementation of the field photography and the execution of the sitespecific weed control technology. The autonomous robot can perform a wide and diverse range of monitoring tasks, because the control unit can be reprogrammed according to the purpose. 3.2. Relationship between NDVI indexes and weed coverage in winter wheat The site-specific nitrogen top dressing of winter cereals based on measurement of biomass and various vegetation indices is spreading and increasing. The means of measuring the vegetation index (Yara N-Sensor, Trimble GreenSeeker etc.) are not able to distinguish between the cultivated plant and weeds. In winter cereals, due to mild autumn and winter weather, many weeds from the T1 and T2 life-form group may be germinating. These species can form high-mass patches (Stellaria media, Veronica hederifolia). In our two-year experiments, we researched the impact of these weed species on the NDVI vegetation index. During our investigations we have 8
found significantly different vegetation indexes in weed-free and weedy cereals. The significant differences were also mathematically verified. If the remote measurement of the NDVI vegetation index is not combined with a weed survey, we may receive false results regarding the adequate nitrogen doses. To avoid these problems, we recommended the application of an autumn weed-control technology in winter-sown cereals. 3.3. Weed control experiment in organic grown sunflowers Our research has been motivated by increasingly widespread organic and chemical-free cultivation. As is known for row-crops (sunflower, corn) at the initial stage of first two months of development a strong weed pressure can be expected. These plants cannot be cultivated successfully without treating this problem. Tests have been conducted with precision inter-row cultivator, and precision inter-row cultivator fitted with rotary finger weeders. The work of the two instruments were evaluated separately. The statistical assessment stated that the sensor-driven cultivators equipped with finger weeders provided sufficiently effective results in organic sunflower. The effectiveness of our developed method was confirmed by the yield of 2.9 tonnes per hectare, which even exceeded the national average harvest of 2.83 t/ha produced in conventional technology with herbicides. However, the price of organic sunflower is comparable to the price of seeds, but the cost of inputs is a fraction of the amount of the conventional technology. As a result, a minimum a double income can be realised in the most modest calculations, without environmental pressures. We have pointed attention to the fact that the verification of certain details of the new technology requires special attention from the managing 9
agronomist and the operator of the machinery. We have shown the possible improvements in technology development. 3.4. Effects of site-specific weed control based on sensor technology on changes of the weed flora of arable lands The results of 9 years of study and research in autumn cereals, 51% of herbicide savings were achieved by the site-specific method used. At the same time, the occurrence of weeds in the treated fields decreased. From our studies in row-crops the chemical-free solutions do not give a comforting result against the perennial weeds (G life-form group) but significantly reduce the presence of Ambrosia artemisiifolia (L.). In the untreated areas, an average of 201 weeds /m 2 population has been reported, which has been reduced to 34.5 weeds/m 2 with a single treatment of the precision inter-row cultivator + rotary finger weeder. The developed method worked effectively against the competitive pressure of weeds, which is also confirmed by the crop yield. The fact, that there has been a short time since the introduction of the precision weed control technology makes it difficult to justify the results, so although long-term results are encouraging, they can be confirmed later. 10
4. NEW SCIENTIFIC RESULTS (THESIS) 1. We have created an off-line site-specific weed control process with digital elements in winter cereals, from weed-mapping to the implementation of the site- specific weed management technology and the autonomous weed-mapping robot is integral part of it. 2. It has been found that in the case of cereals in early development stages, weeds in the area may result significant errors in the determination of NDVI vegetation indices of crops. The adequate amount of site-specific nitrogen can be obtained by using remote sensing data collected only from weed-free fields. 3. We have developed a herbicide-free weed control technology based on mechanical implements controlled by sensor technology, and one of the basic conditions is the usage of a precision sensor-guided inter-row cultivator equipped with rotary finger weeders. 11
5. PUBLICATIONS RELATED TO THE SUBJECT OF THE PRESENT DISSERTATION ARTICLES: BODON D, REISINGER P ÉS BORSICZKY I, 2009: A parlagfű (Ambrosia artemisiifolia L.) többszöri kaszálásának és glifozáttal történő vegyszeres gyomirtásának hatásvizsgálata. Növényvédelem (8) 440-444 BORSICZKY I, ENZSÖL E, FARKAS B ÉS REISINGER P, 2015: Study of the use of N sensor in weed covered fields of winter wheat. Herbologia Vol.15. No.1 99-109. Academy of Sciences and Arts of Bosnia and Herzegovina. BORSICZKY I, ENZSÖL E, FARKAS B ÉS REISINGER P, 2014: Nitrogén (N) szenzorral történt mérések eredményei gyomos és gyommentes őszi búzavetésben. Magyar Gyomkutatás és Technológia (Hungarian Weed Research and Technology) XV. évf. (1-2) 47-57 BORSICZKY I ÉS REISINGER P, 2017: Precíziós gyomszabályozási folyamat szervezése gyomfelvételező robottal, Integrált termesztés a kertészeti és szántóföldi kultúrákban (XXXIV.) 2017. november 23. Budapest 978-963-89690-5-7 REISINGER P AND I BORSICZKY, 2018: From traditional weed mapping to an autonomous robot: developments and results from Hungary,Proceedings 28 th German Conference on Weed Biology and Weed Control, February 27-March 1, 2018, pp. 363-372, DOI 10.5073/jka.2018.458.054 12
BORSICZKY I, ENZSÖL E ÉS REISINGER P, 2018: Gyomszabályozási vizsgálatok eredményei bio-napraforgóban; Magyar Gyomkutatás és Technológia. Megjelenés alatt. BOOK CHAPTERS: REISINGER P, BORSICZKY I ÉS EÖRI T, 2012: Repceteremesztés 45 cm-es sortávolságra, mulcsos technológiával. 143-151 (In: Eöry T.: Versenyképes repcetermesztés. Mezőgazda Kiadó. ISBN:978-963-667-3.) BORSICZKY I, 2016: Optikai szenzoros vezérlés a sorközművelésben, (In MILICS G (szerk.) II. PREGA, PRECÍZIÓS GAZDÁLKODÁS, Digitalizáción innen és túl p.53.); ISBN 978-963-12-5342-9 BORSICZKY I, ENZSÖL E, REISINGER P, 2017: Nitrogén (N) szenzorral történt mérések eredményei gyomos és gyommentes őszibúza vetésben, (In MILICS G (szerk.) III. PREGA, PRECÍZIÓS GAZDÁLKODÁS, PREGA Science Konferencia p. 102.); ISBN 978-963-12-8921-3 CONFERENCE PRESENTATIONS AND ABSTRACTS: BORSICZKY I ÉS REISINGER P, 2018: Gyomszabályozási vizsgálatok eredményei bio-napraforgóban, A Dr. Ujvárosi Miklós Alapítvány a gyommentes környezetért 35. találkozója valamint a Magyar Gyomkutató Társaság 24. Konferenciája, 2018.március 08., Nova BORSICZKY I ÉS REISINGER P, 2017: Gyomszabályozási vizsgálatok eredményei bio napraforgóban. XXX. Biokultúra Tudományos Nap. 2017. december 2. Budapest 13
BORSICZKY I AND P REISINGER, 2017: Demonstration of an autonomous photo-optical weed mapper, Joint EWRS Workshop of the Working Groups Physical and Cultural Weed Control and Crop-Weed Interactions, Nyon, Switzerland, 2-5 April 2017 BORSICZKY I, REISINGER P, 2016: Gyomszabályozási módszerek az ökogazdálkodásban, 2016. dec. 3. Biokultúra Tudományos Nap BORSICZKY I, 2014: Differenciált kezelési zónák kijelölése elektromágneses talaj-szkenner segítségével a precíziós gazdálkodás gyakorlatában. A talajok térbeli változatossága - elméleti és gyakorlati vonatkozások. ea. Talajtani Vándorgyűlés Keszthely, 2014. szeptember 4-6. BODON D, REISINGER P ÉS BORSICZKY I, 2009: A parlagfű (Ambrosia artemisiifolia) többszöri kaszálásának és glifozáttal történő vegyszeres gyomirtásának hatásvizsgálata. 55. Növényvédelmi Tudományos Napok. Budapest. p 45. BORSICZKY I, HOFFMANNÉ PATHY ZS ÉS REISINGER P, 2008: Precíziós gyomszabályozás Weed Seeker-rel, kukoricában, Óvári Tudományos Napok, Konferencia kiadvány OTHER PUBLICATIONS: REISINGER P, BORSICZKY I, 2009: Precíziós gyomszabályozás Gyomvadász intelligens szórófejjel. Agrofórum Extra 27. 68-70. BORSICZKY I, REISINGER P, 2013: Precíziós megoldások a gyomnövények ellen. Biokultúra. 2013/2. 32-33. REISINGER P, BORSICZKY I, 2013: Precíziós növényvédelem gyomszabályozás. Agro Napló 2013/5. 66-68. 14
BORSICZKY I, REISINGER P ÉS TARSOLY K, (2017): Gyomok elleni védekezés kenőgéppel, cukorrépában, Agrárágazat, 2017. október, pp. 50-51. BORSICZKY I, REISINGER P ÉS TARSOLY K, (2017): Gyomok elleni védekezés kenőgéppel, cukorrépában, Cukorrépa, XXXV évf. 2017/1, pp 20-21 BORSICZKY I, 2017: A precíziós gazdálkodás gépesítése, (In Borsiczky et al., A precíziós gazdálkodás lehetőségei Szerbiában pp.21-23); ISBN 978-86-89827-09-5 PATENT: BORSICZKY I, REISINGER P, PATKÓ G, 2017: Egyedi fejlesztésű, használati mintaoltalommal rendelkező autonóm (önjáró) gyomfelvételező eszköz, (Eszköz kezelendő terület precíziós növényvédelmének megkönnyítésére, Szellemi Tulajdon Nemzeti Hivatala, használati minta NSZO-jelzetei: G02B 27/00 A01M 7/00 A01M 21/00) 15