Embed Size (px)
Citation preview
application and reduction of pesticides di stri -bution. These traits could be monitored using toolsthat can describe the plant status and its envi -ronmental development in detail. The scenario isforeseen where the farm will become a multi -functional subject, productive but also environmen-tally sustainable.Agricultural practices generate strong pressures onthe environment, particularly with regard to changesin habitat, land and water use, and emissions ofhazardous inputs (UNEP - International ResourcePanel 2010). Indeed, 70% of freshwater withdrawalsare used for agriculture; in the coming years we willface the challenge of producing more food with lessuse of water resources. In addition, according to thelatest assessment by the Intergovernmental Panelon Climate Change (IPCC - Fifth AssessmentReport - 2012), agriculture is responsible for over13.5% of greenhouse gas emissions, a value higherthan that reported for the transport sector (13%).Agriculture also contributes significantly withsynthetic chemicals (pesticides). At the same time,
Agrometeorological monitoring: Low-Cost and Open-Source – is it possible? Alessandro Matese1*, Salvatore Filippo Di Gennaro1-2, Alessandro Zaldei1
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
81
1. INTRODUCTIONAgricultural systems are in general highlyheterogeneous, due either to intrinsic factors suchas topography and soil characteristics or externalfactors such as cropping practices and seasonalweather trends. Precision farming was establishedwith the aim of managing crop system variabilityin order to optimize agronomic practices,rationalize the use of inputs and maximize quality.Sustainable agriculture must be characterized bythe following distinctive features: (i) reduction ofenvironmental impacts, (ii) reduction of costs and(iii) maintenance of high quality yields. Among thechallenges faced by decision-makers, an under-standing of the crop nutritional deficit and waterstatus plays a key role for variable rate fertilizer
Abstract: The paper describes the development and application of a low-cost and open-source system foragrometeorological monitoring. The system is based on the Arduino platform and agrometeorological sensors. Data aretransmitted using a WiFi Shield and GSM/GPRS Shield. The system is equipped with different agrometeorologicalsensors: air temperature and relative humidity, solar radiation, wind speed, rain gauge, leaf wetness, atmosphericpressure, soil moisture, and a solar panel and battery made it energy self-sufficient. After a testing period in the lab,the field test was conducted in an experimental vineyard of CNR - IBIMET in the summer of 2013. The system hasprovided excellent results in terms of accuracy and stability of the acquired data from each sensor. This study showsthat it is possible to monitor agricultural systems with low-cost devices. The potential of the system is high, as it hasproved to be highly flexible to the different needs of the user due to the open-source philosophy, allowing maximumcustomization in terms of programming and the possibility of adding a wide range of sensors.Keywords: Arduino, agrometeorological monitoring, precision farming, sensors, low-cost, open-source.
Riassunto: Il presente lavoro descrive lo sviluppo e l’applicazione di un sistema a basso costo e con software libero peril monitoraggio agrometeorologico. Il sistema è basato sulla piattaforma Arduino ed un set di sensori per il monitoraggioagrometeorologico in agricoltura. La trasmissione dati è stata realizzata equipaggiando la piattaforma con specificimoduli Wifi e GSM/GPRS. I parametri monitorati sono stati la temperatura e umidità relativa dell’aria, la radiazionesolare, la velocità del vento, le precipitazioni, la bagnatura fogliare, pressione atmosferica e umidità del suolo. Il sistemaera autonomo da punto di vista energetico perché dotato di un pannello solare e di una batteria tampone. Dopo unperiodo di test in laboratorio, il sistema è stato testato in un vigneto sperimentale del CNR-IBIMET e ha mostratoottimi risultati in termini di flessibilità di utilizzo, qualità dei dati acquisiti e trasmissione dello stesso. Il sistema hadimostrato di poter utilizzare sistemi low cost per il monitoraggio in agricoltura e la flessibilità data dall’utilizzo disoftware libero che permette la personalizzazione rapida per un maggior ventaglio di possibili utilizzi.Parole chiave: Arduino, monitoraggio, agrometeorologia, agricoltura di precisione, sensori, basso costo, softwarelibero.
*Corresponding Author e-mail: [email protected] IBIMET CNR – Istituto di Biometeorologia, Consiglio Nazionaledelle Ricerche, Firenze (FI).2 DSAA - Dipartimento di Scienze Agrarie e Ambientali, Universitàdi Perugia, Perugia (PG).
nota tecnica
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 81
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
82
satisfy observational and statistical requirements,and this can quickly become cost-prohibitive. Theopen source philosophy represents a breakthroughof considerable significance in the history ofknowledge and scientific research. Open-source isbased on an intellectual exchange and recognitionof copyright in development mode, which ispromptly reinvested to update and improve theproject.In recent years, the advent of low-cost micro -controllers has led to a rapid use in the scientificcommunity (Vellidis et al., 2008; Fisher and Kebede,2010). The possibilities afforded by an open-sourcehardware system, the most famous example being theArduino project (Arduino, 2012), include the rapidprototyping of ICT systems where circuits models arelicensed under Creative Commons and can bemodified by the user. This leads to a coordinateddevelopment of the hardware and software with largenetwork communities, which provide effectivesupport. An advantage of open-source hardware isthat a wide variety of ready-to-use – with fewmodifications – software is available for them on theWeb, shortening development times. A wide rangeof low-cost interfaces, accessories and sensors arealso available on the Internet, along with usefulinstructions.The objective of this work was the development oftwo low-cost and open-source prototype agrometeo -rological stations for monitoring agrometeo-rological parameters.
2. MATERIALS AND METHODS
2.1. System descriptionThe developed system (Fig. 1) is based on theArduino platform equipped with an 8-bitprogrammable microcontroller ATmega328(Atmel Corporation, San Jose,CA USA). Themicro controller contains 32 kilobytes (KB) offlash memory for program storage and 1 KB ofnon-volatile memory data storage. The I/O linesare made up of various digital and analog pinswith 10-bit resolution. The device operates at 5V16 MHz or 3.3V 8 MHz. The Arduino board isdesigned to allow for expansion through theconnection of auxiliary boards or shields. Theshields are connected through connecting pinsthat are arranged in the same physicalconfiguration as the Arduino board. Programminglibraries enable users to rapidly integrate newdevices and sensors into projects without writingan extensive new program routine. TheSeeeduino Stalker v2 is a hardware platform
the agribusiness companies have to satisfy the needsof consumers, who are increasingly aware of foodsafety and quality standards. In this context,particular attention is placed on local products thatcharacterize the “Made in Italy” agro-food, which isoften associated with historical-cultural valuesrelated to production areas within a vision thatcombines food and territory. Taken together, theseconsiderations highlight the need for a researcheffort and innovation to find solutions to limit andrationalize the use of resources and minimize theuse of chemical fertilizers. To pursue theseobjectives, integrated tools have been developed forcrop monitoring. Among the agrometeorologicaldevices, the Wireless Sensor Networks (WSNs) aretaking the spotlight. The WSNs are networks ofsensors located within the crop systems that canmeasure, collect, process and share data in real timeon the Web, making them immediately accessibleto the end user. The spatial geometry of thesenetworks is of great interest for applications inagriculture, by identifying conditions at differentscales (Matese et al., 2014), which enable thedifferent micro-areas to be pointed out within acropping system. Indeed, many micrometeoro-logical parameters play a crucial role on the planthealth status and physiology. This new technologicalscenario allows the farmer to manage informationand thus to know in real time the detailed conditionsin the field, thanks to the combined use of cutting-edge sensors, instant data transmission and theiraccessibility on various devices. It becomes possibleto program interventions such as irrigation or plantprotection treatments, with advantages that are notonly economic but also in terms of sustainability.The paradigm of “low-cost” and the integration ofDSS (Decision Support System) data flows fromWSN, will make it possible to directly implementintervention strategies/responses in the field.Research in agrometeorology requires the moni-toring of a large amount of data and a compromiseusually has to be made between the type andamount of measurements required and theresources available for collecting. A large numberof electronic solutions are available for automaticmonitoring of experimental fields. Often monitoringsystems contain proprietary technology thatmanufacturers do not wish to release, and arefrequently designed to operate with only a particularmanufacturer’s sensors. The users are locked intoparticular systems and technologies with steep costsfor the various types of research they intend toundertake. In addition, scientific experimentsrequire multiple sites and replicated treatments to
nota
tec
nica
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 82
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
83
nota tecnica
The system (Fig. 2) involved the use of two differentdata transmission modules, one based on GPRScommunication using a Seedstudio GPRS Shield V2.0modem and the other using a WiFi module(Seeedstudio WiFi Shield v1.1). With the GPRSmodem, it is possible to use Arduino to dial a phonenumber or send a text message using AT commands.The modem has a quad-band low-power consumptionwith a GSM / GPRS module SIM900 and compactantenna PCB and is compatible with all types ofArduino. WiFi Shield uses a WiFi module RN171 toprovide Arduino Ethernet standard functions. It ispossible to easily connect the device to the wirelessnetwork 802.11b/g. With support for common TCP,UDP and FTP communication protocols, this WiFiShield can meet the wireless network needs of mostprojects. It is also compatible with all types of Arduino,has a secure WiFi authentication WEP -128 , WPA-PSK ( TKIP) , WPA2- PSK ( AES), and many built-innetworking applications DHCP client, DNS client,ARP, ICMP ping, FTP, TELNET, HTTP, UDP, TCP.
based on a programmable microcontrollerATmega328 8-bit and designed for wireless sensornetworks. It is completely Arduino compatible,but integrates the devices that make it usefulindividually without the use of a few shields: oneslot for Micro SD card, 20 I/O pins, com -munication protocol I 2 C and UART, wirelessmodules connector for Xbee components such asZigBee, Bluetooth and RF. It also includes asocket for connecting a power supply (battery)and a socket for connecting a solar panel, usefulfor charging the battery. The Seeeduino boarddoes not have a USB-to-serial converter chip on-board, in order to reduce costs and energyconsumption. A special USB-to-serial converter,FTDI cable, is used as an interface to programthe card and create a virtual serial port.The software environment for programming withArduino is available for download and installationin various computer operating systems (GNU /Linux, Mac OS X and Windows). Using the IDE,the user writes programs in a language based onC++. The IDE then compiles errors anddownloads the compiled routine to themicrocontroller. A terminal window is availablefor the text output from the Arduino board to thecomputer serial monitor. The contribution of thecommunity is fundamental because, being anopen-source project, it is possible to find manyprogramming libraries that contain routines tosimplify programming and incorporate advancedfeatures, sample code and complete programs areavailable to download, use and modify, ifnecessary.
Fig. 1 - Developed system.Fig. 1 - Sistema sviluppato.
Fig. 2 - Hardware components.Fig. 2 - Hardware del sistema.
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 83
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
84
operational (5μA) in sleep-mode (1μA), combinedwith low cost (about 10 Euros) make this productexcellent for monitoring needs.
2.2.3. Solar radiationThe system is equipped with a sensor for incidentsolar radiation measurement. The technology isbased on a photodiode contained in a Teflonstructure to measure the diffuse component of thesolar radiation (Fig. 4). Among the different modelsa silicon photodiode (Hamamatsu, S1226 -BK) waschosen in order to ensure a spectral response in thewavelength range 200 – 1000 nm. The sensorrecognizes a specific wavelength of the electroma-gnetic wave incident and transforms this into anelectric current signal by applying an appropriateelectric potential to its ends. The advantage of thissensor over a photoresistor is the very high responsespeed, and it also provides a noise-free output signalwith excellent linearity with respect to the incidentlight. This sensor perfectly fits the needs of thesystem due to the strength and longevity of thesensor, plus the fact that the operation of thephotodiode does not require a power supply, butmost especially because of a very good ratiobetween data quality and cost (5 Euros)
2.2.4. Leaf wetnessThe Leaf Wetness Sensor detects the presence ofsurface moisture. This sensor works by having aseries of exposed traces connected to the groundand the sensor traces are interlaced between the
2.2. Sensor equipmentThe devices allow the connection of a large numberof sensors. The sensors operate at low voltages andoutput signals compatible with the microcontroller,including analog voltage, variable frequency and aselection of digital communication protocols. Forthis application, the following agrometeorologicalsensors have been used:
2.2.1. Air temperature and humidityA DHT22 sensor (Sensirion AG, Switzerland) wasintegrated into the system for measuring the airtemperature and relative humidity (Fig. 3). Thesensor module, thanks to the very small dimensionsof the CMOSens® technology, provides a digitalsignal for fully-calibrated relative humidity (±.5%)and temperature (0.5 °C) allowing full integrationinto the system and excellent long-term stability.Further advantages are the very short response timeof 4 seconds and the wide range of operatingtemperatures from -40 to 120 °C. The cost (10Euros) as well as power consumption (up to 1μW)are minimal, this makes it a perfect sensor for low-cost and low-power applications.
2.2.2. Atmospheric pressureAtmospheric pressure is measured by an MPL115A1sensor (Freescale Semiconductor, Inc.). The digitalbarometric sensor uses MEMs technology to providean accurate measurement within a range from 50kPato 115kPa. The measurement accuracy, small size (5 x 3 x 1.2 mm) and minimum power consumption in
nota
tec
nica
Fig. 3 - Air temperatureand humidity sensorinstalled in the field.Fig. 3 - Sensore di temperatura ed umiditàdell’aria installatoin campo.
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 84
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
85
cost (10 Euros). Furthermore, the sensitive metalpart ensures a minimum inertia of the measure(response time < 750 ms), good precision (0.5 °C)and high resolution (from 9 to 12 bits).
2.2.7. Wind speed and Rain gauge To measure the wind speed and precipitation a low-cost (76 Euros) Weather Sensor Assembly p/n80422 by Argent Data Systems was used (Fig.5).Therain gauge is a self-emptying tipping bucket type.Each 0.011” (0.2794 mm) of rain causes onemomentary contact closure that can be recordedwith a digital counter or microcontroller interruptinput. The cup-type anemometer measures windspeed by closing a contact as a magnet moves past aswitch. A wind speed of 1.492 MPH (2.4km/h)causes the switch to close once per second.
2.3. Power supply The power supply consists of 12 V/ 4.5 Ah leadbattery (Fig. 1) and a 50Wp solar panel (Fig. 5). A 5V low-dropout DC/DC converter has been addedto adjust the battery voltage (Fig. 1). The elementsare contained in a watertight case with IP 67protection.
3. RESULTS AND DISCUSSIONThe agrometeorological application implementedtwo systems that differ by sensors equipment. Thefirst system is a traditional agrometeorological
grounded traces. The sensor traces have a weakpull-up resistor of 1MΩ. The resistor will pull thesensor trace value high until a drop of water shortsthe sensor trace to the grounded trace. This circuitworks with the analog pins of the board and detectsthe amount of water-induced contact between thegrounded and sensor traces. The price is around 20Euros.
2.2.5 Soil MoistureSoil moisture has been monitored by equipping thesystem with a WaterScout SM100 sensor (SpectrumTechnologies, Inc.). The sensor is based on theFDR (Frequency Domain Response) technique,which determines the volumetric percentage ofwater in the soil (WVC) measuring the change inthe dielectric constant by means of a radiofrequency electronic circuit. It presents excellentperformance in terms of high-resolution (0.1%) andaccuracy (3%) without the need for calibration. Theprice is around 100 Euros.
2.2.6. Soil temperatureSoil temperature is measured by a DS18B20 sensor(Maxim Integrated TM, California, USA). It is awaterproof digital probe, with the body coveredwith plastic film PVC insulation and the sensitivepart in stainless steel, which makes it suitable forany gaseous, solid or liquid measuring environment.This sensor was chosen for its versatility and low
nota tecnicaFig. 4 - Solar radiationsensor installed in the field.Fig. 4 - Sensore di radiazione solareinstallato in campo.
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 85
station with the sensors described in materials andmethods. The second system can be consideredas a low-power node without rain gauge oranemometer that must be handled with interruptrequiring a fixed idle state of the board. The lattersystem was planned with a “sleep mode” routinein which consumption is considerably reduced,only a few mA on average. A comparison versusresearch grade agrometeorological stations couldbe reported considering Campbell Scientific andDavis weather stations as typical systems used forresearch purposes. A CR1000-based weatherstation, measuring standard meteorologicalsensors, and Vintage Pro 2 have an averagecurrent drain of 0.8 mA. The setup was verysimple thanks to the “sandwich” implementationwith several layers (shields). The libraries for themanagement of all sensors were easily found onthe Web and then modified. The softwareconsisted of a three layer script: (i) a root programwith the library initialization and time-elapse fordata acquisition, (ii) a second program with thesensor acquisition procedures and (iii) the thirdwith the data transmission procedures byGSM/GPRS or via the WiFi module. With regardto the data format restitution, since these systemshave been used mainly for research purposes,only string values were transmitted and stored.On the server side, we had a Linux CentOS 5server running on a i686 machine, with aVSFTPD server set-up. Output data were stored
with hourly frequency, and a simple controlprogram was compiled to check dataset consi -stency, and notify the operator of any problemthat might have arisen in the transmission chain. The firmware provided the data acquisition andtransmission from the sensors every 15 minutes,once acquired from the remote server; the datawere stored in “csv” format files. After a testingperiod in the lab, the field test was carried out in aCNR - IBIMET experimental vineyard (SestoFiorentino, Italy) in the summer of 2013. Thesystem has provided excellent results in terms ofaccuracy and stability of the acquired data fromeach sensor. The power supply test in the laboratoryallowed us to evaluate an efficiency of at least 10days of energy independence in the absence of solarpanel charging. This confirms the great potential ofthe system in terms of both low energy consumptionand high operational efficiency. In detail, theexperience gained from these tests has shownadvantages and limitations in the use of open-sourcetechnologies and low-cost devices for agrome-teorological applications:As regards the advantages:• The quality of the electronic components of thesedevices is already good and reliable for the purposesof agrometeorological monitoring;• The prototype cost including devices and relatedArduino shields, sensors, box and all components forpower supply and installation is estimated at around500 Euros. This allows for a dense monitoring useful
86
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
nota
tec
nica Fig. 5 - Wind speed,
rain gauge and solar panel.Fig. 5 - Anemometro,pluviometro e pannellosolare.
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 86
for many applications, where the data are usuallytaken from a small number of stations that are notrepresentative of the experimental site;• The Arduino community (http://forum.arduino.cc)shares all the libraries, programs and hardwarescheme, so the user could easily find a solution toall kinds of issues on the GitHub database.• The Arduino based technologies enable rapidprototyping and use of a wide range of sensorsuseful for all the different monitoring needs.Concerning the limitations:• The low-cost sensors quality does not conform tostandards set by WMO (World MeteorologicalOrganization), so a characterization and calibrationstudy becomes necessary of each sensor’s responseswith reference to WMO standards sensors. Figure6a-b show the comparison results of low cost versusreference sensors equipment in laboratory. The lowcost temperature and humidity sensor wascompared with a Vaisala HMP45 humidity andtemperature probe as reference (Fig 6a). Solarradiation sensor was compared with a HuksefluxSR11 pyranometer (Fig 6b);• Traditional equipment have a much larger memorycapacity for data-logging purpose compared to lowcost devices.• The hardware elements setup is still at theprototype stage, which presents problems inindustrialization and large-scale production;• Open-source community could suffer a negativeeffect due to the loss of control or standards;• One of a limiting factors to the diffusion of open-source technology is that companies show lack ofinterest in a non-proprietary product, because itcould be easily copied.• The research experience in weather monitoringprocedures problems plays a key role in drivingtechnological innovation in an open-source andespecially low-cost direction.Many studies have presented monitoring solutionswith the prerogatives of open-source and low-costfor different applications, for example the controlof soil moisture and field irrigation (Fisher andGould, 2012) or biomass evaluation using opticalsensors for image acquisition (Kanda et. al., 2011).Others proposed wireless sensor applications inprecision viticulture (Matese et al., 2013; Zacha-riadis and Kaskalis, 2012), which enable site-specificmicroclimate monitoring in vineyard. Sensor-weband GeoDB represent the potential of thosetechnologies, and allow information integrationand the creation of interfaces for easy access tosmartphone applications (Apps). At the same timeother solutions that supply an excellent DSS are:
data correlation with location-based information,Internet of Things (IoT) applications, models forthe optimization of sustainability in agronomicinterventions, implementation of an actuators-based system (e.g. management of automatedirrigation) and data analysis providing semanticaggregation.
4. CONCLUSIONTechnological progress has created a fertile substratefor a new phase of agricultural moderni zation, interms of low-cost, adaptability and flexibility, makingit more accessible for scientific purposes andoperating innovative tools aimed at supportingdecision-making. Although the system presented inthis work is an early-stage prototype, the preliminaryresults have shown good per formance in terms ofautonomy and quality of data transmission, withfeatures that meet any re quirement of microme -teorological monitoring, such as power consumption
87
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
Fig. 6a-b - Comparison results of low cost versus referencesensors equipment in laboratory.Fig. 6a-b - Confronto tra sensori low cost e sensori di riferimentoin laboratorio.
a
b
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 87
and low cost. The potential of the system is high, asit has proved to be highly flexible to the differentneeds of the user due to the open-source philosophy,allowing maximum customization in terms ofprogramming and the possibility of adding a widerange of sensors. Moreover, it suggests the need forevery technological institute to provide aninstitutional “Github” where it shares all the codesdeveloped for applications and not just publishscientific papers in which very often the “experimentrepeatability” is, in our opinion, very ephemeral.Going back to the initial question: Agrometeoro-logical monitoring: Low-Cost and Open-Source - is itpossible? The answer we can give based on ourexperience is: WE CAN DO IT!
5. ACKNOWLEDGEMENTSA special thanks to Francesco Sabatini, JacopoPrimicerio, Tiziana De Filippis, Leandro Rocchiand Alfonso Crisci from IBIMET-CNR.
REFERENCESArduino 2012. An Open-Source Electronics Proto-
typing Platform. http://www.arduino.ccFisher D. K. and Kebede H., 2010. A low-cost
microcontroller-based system to monitor croptemperature and water status. Computers andElectronics in Agriculture. 74: 168-173.
Fisher D.K. and Gould P.J., 2012. Open-sourcehardware is a low-cost alternative for scientificinstrumentation and research. Modern Instru-mentation. 1: 8-20.
Kanda K., Ishii T., Kameoka T., Saitoh K., Sugano R.,2011. Field Monitoring System Using Agri-Server.SICE Annual Conference 2011, September 13-18,2011, Waseda University, Tokyo, Japan
Matese A., Crisci A., Di Gennaro S. F., PrimicerioJ., Tomasi D., Marcuzzo P. and Silvia Guidoni,2014. Spatial variability of meteorologicalconditions at different scales in viticulture.Agricultural and Forest Meteorology. 189-190:159-167.
Matese A., Vaccari F. P., Tomasi D., Gennaro S. F.D., Primicerio J., Sabatini F. and Guidoni S.,2013. Crossvit: Enhancing canopy monitoringmanagement practices in viticulture. Sensors(Switzerland). 13(6): 7652-7667.
Vellidis G., Tucker M., Perry C., Kvien C. andBednarz C., 2008. A real-time wireless smartsensor array for scheduling irrigation. Computersand Electronics in Agriculture. 61: 44-50.
Zachariadis S. and Kaskalis T. H., 2012. An EmbeddedSystem for Smart Vineyard Agriculture. 2nd Pan-Hellenic Conference on Electronics and Telecom-munications - PACET’12, March 16-18, 2012,Thessaloniki, Greece.
88
Ital
ian
Jour
nal o
f Agr
omet
eoro
logy
- 3/
2015
Riv
ista
Ita
liana
di A
grom
eteo
rolo
gia
- 3/2
015
nota
tec
nica
81-88 NT matese:Layout 1 19-01-2016 12:38 Pagina 88
