Final Report Summary - RICE-GUARD (In-field wireless sensor network to predict rice blast)
Executive Summary:
The RICE-GUARD project developed an on-line, wireless sensor network in-field environmental monitoring system capable of measuring several environmental parameters (temperature, relative humidity, solar radiation, leaf wetness, among others) from inside the rice canopy. This system provides unprecedented data from different levels within the rice plants and is capable of substituting data logger nodes located in weather stations outside the plants (where the conditions are much different than inside the canopy) and single parameter measuring hand-held meters used when collecting accurate data inside the rice areas. This information could be used to model related factors affecting the crops such as the presence of pests, fungal diseases or other plant pathogens. In this regard, in combination with the hardware prototype developed, the RICE-GUARD system has also developed a tailored model for determining the risk levels of rice-blast (Pyricularia oryzae) incidence. One of the most important features of the system is its ability to provide remote readings of the environmental condition within the rice, as well as the level of risk associated to those, in a timely manner. Rice-growers have been able to read their local, in-field conditions using a tailored user interface which graphically represented the level of susceptibility to rice blast under which their rice fields were growing. This can result in a ultimate optimisation of the management of rice crops, ensuring a rational and efficient use of the commonly used, and potentially harmful, fungicides and chemical protective substances.
The RICE-GUARD system is composed by autonomous nodes containing 4 levels of temperature, relative humidity, leaf wetness and solar irradiance sensors, a communications module and an energy harvesting unit made of solar panels; an autonomous master node, equipped with all the sensors commonly found in weather stations plus a communication system receiving the readings of all nodes in the area and storing the information on a local database while uploading it to the cloud; and a user interface displaying the information coming from the field and applying the rice blast prognosis model.
RICE-GUARD was validated in 5 rice cultivation areas in Europe and proved to provide valuable information to all of them. It was installed in rice paddies under controlled and non-controlled conditions in order to prove the adaptability of the model. This allowed to effectively develop a model that was easily adaptable to all areas and which was demonstrated to predict the risk conditions in which the crops where infected. This was reinforced by the positive cross check of data with another European project, ERMES.
It was demonstrated that the RICE-GUARD system is able to provide much more accurate information than sensors located near fields but not inside them while allowing a real time, remote and user friendly monitoring of this information. This functionality was proven to work for managing rice blast incidence in rice crops and therefore increasing the yield, optimizing the physiological characteristics of rice and producing a safer food product.
The RICE-GUARD system that led to a route to market plan will need further validation, especially on the model testing, since creating robust models for reliable agricultural decision making takes usually between 5-6 years.
Project Context and Objectives:
Rice is the main crop in wet areas such as river deltas and is an essential tool in Europe in managing protected ecosystems. Irrigation water is a key factor in the production of rice and water quality has a major impact on crop yield as a result of tolerance of rice to factors such as dissolved salts. Rice is more water consuming than many other crops; in continuous flooding cultivation it takes about six times the water required by wheat. Due to increased water use in coastal areas, the sea intrudes the water table and rivers, and seawater floods nearby fields during storms in the Mediterranean area. The result is increased water salinity, which reduces yield in rice crops and increases soil salinity. Nowadays, water condition is for the most part assessed by visual inspection of the crops and, when excess water salinity is suspected, fields are irrigated by flooding them. In areas where water salinity is endemic, rice paddies are continuously irrigated with river water to reduce water salinity. This is a remedial solution that requires enormous volumes of water and considerable energy to pump water.
Water salinity can be accurately determined by measuring its electrical conductivity (EC). Measuring EC at the water inlet and outlet of each paddy field can help in monitoring the “washing” effect of irrigation. Moreover, measuring EC at points far from water inlets and outlets can help in assessing water salinity in a given paddy field and at different depths in drainage channels can help in managing water salinity in larger areas.
Based on this background the RICE-GUARD project built upon the firstly developed prototype system based on a master node (MN) and a series of paddy nodes (PN) and repeater nodes (RN) to validate the fabrication of the final RICE-GUARD system. On the other hand, the system was used to obtain the data needed for the rice blast prediction model refinement during the second rice season of the project. Two complete RICE-GUARD systems were installed and tested in Greece and Seville, while series of autonomous data-logging PN were installed in Portugal (2), Italy (1), Turkey (1) and Greece (3).
The main objective of RICE-GUARD was to obtain local information than can improve the accuracy and timeliness of existing rice blast forecasting models that currently rely on information sometimes obtained at more than 50 km from rice fields.
Firstly, the needs and specifications of the rice-growing industry were defined to drive the requirements of the different parts of the system. Initial requirements and specifications were analysed and after a series of consultations and after testing with a preliminary system, the final specifications were obtained. For the paddy nodes these included 4 measurement heights, with temperature, relative humidity and solar radiation so that at least 2 of them will be inside the rice canopy even in short varieties. The range of metrological specifications are 5 to 35 ºC for the temperature sensors, 70 to 98% for the relative humidity sensors and 0 to 1500W/m2 for the solar radiation ones. A leaf dew sensor was also deemed necessary due to the importance of the dew appearance which stays on the plant leaves. The nodes should be solar powered and with a communication system operating in the unlicensed ISM band at 433MHz. The master node should be able to work as a weather station all year long and should include temperature (-10 to 55 ºC), relative humidity (5 to 100%), solar radiation (1 to 1500 W/m2), wind speed (0 to 65 m/s), wind direction (0 to 360 º), rain precipitation (0 to 125 mm/h) and atmospheric pressure (660 to 1070hPa) sensors. It also should be prepared for mains power connection as well as solar powered and prepared for 3G or Wi-Fi connectivity. The user interface should show graphs on the mean readings in the field, graphs for one specific node and possibility to select periods of time to be graphically represented.
The sensors, their power supply system and the in-field user interfaces were developed following the designs and verification tests performed at laboratory scale. A novel remote wake-up system for the Bluetooth interface has been designed and successfully tested, achieving up to 22m communication range, significantly beyond the expected 10m. Similarly, very good results have been achieved in reducing the power consumption of the master node supply and sensor unit (SSU) and size (50%) compared to the preliminary Compact Weather Station system, despite the extra leaf wetness sensors.
An on-line wireless system for the measurement of salinity was designed and developed based on the requirements defined initially. This wireless sensor network (WSN) comprised independent nodes with Radio Frequency (RF) technology that communicated with a gateway device which gathered all data coming from the sensors and stored it locally and in a web server. It also included repeater nodes in order to cover a longer distance in large rice fields. The WSN was optimised in terms of distance of communications, message routing protocol and power consumption to meet the overall requirements of the system. Moreover, several autonomous data logger paddy nodes were designed and included temperature, relative humidity and leaf wetness sensors.
The newly developed sensors and WSN system were integrated into tailored mechanical enclosures which comprised the RICE-GUARD paddy and master nodes. Two new PCBs were designed (one for the communications control and one for the sensors and power supply system) following the research and tests performed previously. The best solution in terms of design simplicity were chosen in order to obtain an easy to assemble and disassemble device. A PVC encapsulation was designed for the nodes which offered the required water proofing capabilities. Similarly, a final PTFE encapsulation solution for the T and RH sensors was found. The system was built to allow laboratory and field testing. An accessible web interface was also developed to show the data obtained in the field in a graphical and understandable way as the front-end of the RICE-GUARD system.
Following a thorough literature review, the Yoshino model for rice blast appearance was decided to be adapted for local rice blast prediction among 50 analysed models. This model granted simplicity and customizability in order to make it widely adopted in all the project regions. The model comprised several parameters such as temperature range, time and wet hours to generate a level of alert directly related to the susceptibility of the plant to become affected by the fungus. After the selection and tailoring of the model, the RICE-GUARD system was tested during two rice seasons to obtain relevant data for enhancing the hardware while refining the model behaviour. Based on these results, recommendations for future improvements to obtain a fully commercial version of the system were outlined.
The process of validation of the RICE-GUARD system took place in 2 locations of the end-users SME-AG partners in the project and involved the testing of the system in rice paddies (Greece and Seville). The custom remote data loggers were tested in the relevant locations for the validation of the model such as rice paddies in Greece, north of Italy and Portugal. This testing at different sites of rice cultivation allowed to evaluate the adaptability of the model and to demonstrate the RICE-GUARD performance and features both within the consortium and beyond to a large number of industry stakeholders from the rice-growing sector.
The RICE-GUARD system proved to be an accurate and reliable tool for the management of rice paddies in rice-blast disease conditions and generally for improving the quality of rice. The close monitoring of the conditions for infection during the cultivation stage allows for a better regulation of the application of fungicides and thus optimises its use and reduces the potential negative environmental effect and at the same time improves food safety.
While developing the RICE-GUARD technology, concern was also given to the cost of selected materials and components to guarantee the future exploitation of the RICE-GUARD system by keeping it of sufficiently low cost to ensure its uptake by the rice growing sector. The developed system cost starts around the 5000€ per complete unit (including gateway, 3 paddy nodes, 1 repeater node and software package) which is in the range accepted by rice growers associations for such a device.
The overriding goal of this project was to ensure that the resulting pre-competitive RICE-GUARD prototype fulfils the set requirements to ensure its further development post-project into a fully industrial system that is taken to market, where its beneficial environmental and socio-economic impact will be felt at European level.
Project Results:
The two main results of the RICE GUARD project have been 1) the demonstration of the remarkable difference between the environmental conditions inside the canopy of inundated rice fields and those above the canopy, not to mention conditions in weather stations far away, and 2) the good prediction capability of the mathematical model developed. These results are relevant because all forecasting models described in the bibliography had been developed from weather data measured outside the canopy, often obtained from public weather stations some kilometres away from rice fields or derived from experiments in greenhouses. The results in RICE GUARD seriously question the scientific validity of studies not based on data measured inside the canopy. Since other fungal diseases in rice plants are also related to the conditions inside the canopy and the correlation between these and the external conditions is very poor in inundated fields, these two results are a leap forward in rice blast combatting and in precision agriculture at large.
These results have required several technological developments that can be applied to other crops and imply a change in paradigm for further developments in plant disease prevention and treatment. In order to describe the specific results achieved, the eight initial scientific and technological objectives of the project are summarized below:
1. To review the technical requirements of the members of the SME-AGs in order to define the system specifications for the RICE-GUARD prototype.
2. To develop weather sensors able to provide the information required by current rice blast forecasting methods that was not available at a local level.
3. To develop the telecommunication system to ensure the effective transmission of data from the RICE-GUARD sensors in-field to a remote data repository.
4. To design the mechanical enclosures for paddy nodes and the master node, to integrate the electronic subsystems, to assemble and test both types of nodes to verify the fulfilment of the electrical, mechanical and metrological specifications.
5. To design a flexible, easy to use, data-driven web-interface and back-end software in order to process the retrieved data, as well as control instructions from operators.
6. To assess weather-based rice blast forecasting models obtained from weather data in other regions in order to customize them by using local weather and additional parameters.
7. To fully test and validate the RICE-GUARD system in rice fields with different environmental characteristics.
8. To meticulously outline post project scaling-up rules and development work for full production.
The research to achieve of the first objective confirmed that no commercial forecasting system neither scientific works are based on real time local weather data obtained within the canopy as offered by a system such as RICE GUARD. The closer approach were measurements with thermo-hygrometer data loggers placed at or close to the border of the rice paddies and manually downloaded weekly, which had been used for some years by one of the SME-AG partners (FAS, Spain) but lacked a scientific validation, i.e. no peer-reviewed scientific publications. Another approach relied on weather data obtained by conventional weather stations inside the rice paddy with sensors above the canopy and some unspecified sensors somewhere within the canopy . The system was part of an extended network wherein data for some regions was extrapolated from measurements at nearby areas.
The research by questionnaires and interviews to farmers demonstrated that they were seriously worried about future regulations that could totally ban the most used fungicide (tricyclazole) and acknowledged the need for forecasting systems with reliable local predictions able to overcome the delay between infection and the appearance of obvious leaf lesions due to pyricularia oryzae. The lack of information about the range, accuracy and resolution of measuring equipment used to derive forecasting models, already detected in the literature review during project conception, as well as about actual environmental conditions within the canopy were confirmed by a more detailed literature review in Work Package 1. Therefore, for each of the three agreed testing sites a compact weather station (CWS) and a moveable field node were designed that together with a handheld node allowed the estimation of the measurement ranges of the environmental parameters involved in the models, their dependence on the position within the canopy, and eventually identify any other factors worth considering when defining the detailed system specifications. The main results of field measurement during the first rice season (2014) were:
1) The large inherent uncertainty of in-field measurands compared to that of the sensors available to measure them meant that there was no need for high-accuracy sensors. Uncertainty for in-field air temperature and relative humidity arises mostly from direct solar radiation and wind hence rice variety, as the height of rice plants ranges from 80 cm to about 120 cm. Commercial instruments having 0.1 °C and 1 %RH resolution seldom yield stable readings in open air, even if there is only light air, in spite of their verified stability inside a climatic chamber in the laboratory.
2) Air temperature and relative humidity inside the canopy are very different from those above it and cannot be inferred from measurements outside the canopy, e.g. those performed by a master node outside rice paddies.
3) Air temperature and relative humidity inside the canopy are determined by water temperature on the one hand and air and relative humidity above the canopy, as well as solar radiation and wind, on the other hand. Measurement at four heights above water level are enough to determine temperature and relative humidity profiles provided at least one measurement point is above the canopy. Measuring water temperature helps in that determination.
4) Air temperature and relative humidity above the canopy can be considered constant only for quite homogeneous areas. Otherwise, it has been verified that data from two public weather stations 10 km apart, one at sea level and the other one about 400 m above sea level can be very different whereas two stations less than 6 km away from each other but both at sea level yield closer readings the same days. Clouds yield fast temperature and relative humidity changes, so that partly cloudy sky results in heterogeneous conditions even in relative small areas that otherwise could be considered homogeneous. Wind speed undergoes large changes even for relatively close measurement points.
5) Since fungicide spraying decisions depend not only from past weather conditions but also from future weather predictions, particularly rain prediction, if no public weather station covers the rice farming area then weather sensors in the master node can be used for that prediction using appropriate meteorological models.
6) Sensors placed within 1 m from the border of a paddy field do not provide representative results of actual canopy conditions. Measurements must performed deeper in the field, which means remote control and data downloading are important to preserve genuine environmental conditions that could be modified by personally accessing the measurement points after installation.
7) Upon seeing the first-year field measurements, industrial partners showed a high interest on the capability of downloading data from in-field sensors even in the absence of Internet access. Consequently, this peer-to-peer communication capability was added to paddy nodes and the master node able to support Android and Windows devices. Both designs were based on off-the-shelf components and that for paddy nodes was particularly compact and low cost, hence offering remote operation capability to any data logger within a range of about 20 m.
The second objective was to develop weather sensors for air temperature (T) and relative humidity (RH), and solar radiation at different heights with respect to ground in paddy nodes placed inside rice fields, and also air temperature and relative humidity, atmospheric pressure, wind speed and direction, rain precipitation and CO2 concentration in master nodes outside the field but close to it. Leaf wetness was detected by calculating dew temperature from air temperature and relative humidity. The correlation between dew detection and leaf wetness detection using commercial sensors was excellent but the method based on T and RH could not detect leaf wetness duration. This raised the need for low-cost leaf wetness sensors.
The major design constraints for paddy node sensors were low cost and power consumption. Commercial integrated T and RH sensors with digital output are inexpensive, but no low-cost pyranometers for solar radiation are marketed. Consequently, it was decided to investigate the use of the very same solar cells used to harvest solar energy to measure the incoming energy flux. This required the solar panel to be split in two halves and measure the respective currents. By carefully orientating each panel, the direct and diffuse solar radiation were correctly measured and the cost of the electronics for power processing was reduced by multiplexing some circuits common to both panels. Field installation, however, was lengthy and needed some skill. Further, any movement of the node that changed the orientation or inclination of the solar cells led to the wrong results. Consequently, whereas the measurement of total solar radiation yields acceptable results, the distinction between direct and diffuse radiation works correctly only for fixed devices.
For the master node, power consumption was less restrictive but cost was still a constraint as usual in industry-oriented research. Therefore, building a weather station by assembling existing meteorological instruments would had been too expensive because of the redundancy of power supplies and communication interfaces of the several instruments required. The approach was to use different commercial sensors and design their electronic interfaces to build a sensor unit (SU) connected to a control and communication unit (CCU) in the same master node. The most cost-effective SU comprised a compact weather sensor (T, RH, barometric pressure, wind speed and direction, and solar irradiance), a CO2 concentration sensor, a rain gage and a leaf wetness sensor.
The third objective was to develop the telecommunication system able to effectively transmit data from the RICE-GUARD in-field sensors to a remote data repository. Initially, a MESH (Multipoint Enhanced Signal Handling) of nodes was expected to be used, but the two planned installations were configured in a linear way requiring a simpler network. The initially adopted solution in the end comprised Radio Frequency (RF) communication between the sensors and the Gateway in the unlicensed 433 MHz ISM band with a multi-hopping message relay protocol between paddy nodes. The selection of this transmission frequency was made according to the price and capabilities criteria. In the market there are several ISM transceivers available which are cheap and at the same they offer important features like high sensibility, low power consumption and programmable transmission power. This satisfied two important requirements of the RICE-GUARD system: it offered low-cost sensor nodes and an autonomous functioning over a long period of time. Moreover, the use of the ISM 433 MHz frequency allowed the implementation of a small-sized antenna that can be easily embedded in the sensor node mechanics.
In the second and third years, a modification to the initial protocol was made although the RF band and technology was maintained. This improvement was implemented to ensure that no data was lost in the multi-hop communication. The new protocol was based on Ad hoc On-Demand Distance Vector (AODV) routing. In this way, sent messages are acknowledged by the receiver and communication only takes place when requested by the master node, improving energy management. Moreover, through these system paddy nodes were able to act as repeater nodes.
The gateway located in the Master Node was the central device gathering all the data coming from the sensors and storing it locally while uploading all information to the cloud server. It was formed by an Alix 3d3 embedded PC system board with an RF transceiver and a USB modem. It used UMTS/GPRS or Wi-Fi protocol to communicate externally with the cloud server.
The fourth objective was to design the mechanical enclosure for paddy nodes and the master node and assemble and test the electronic subsystems before installing them in the field. Three separate designs were created for the Master Node, Paddy Node and Repeater Node.
The master node had all the electronics enclosed in an IP68 plastic box which provided the necessary water tightness for the different PCBs contained in it. The support was based on an off-the-shelf solution adapted for holding the electronics case and the sensors.
The body of the paddy nodes was based on common (non-pressure) PVC cases with IPX3 protection to reduce assembly costs. One case was devoted to the communications and control electronics while another, larger case, was only dedicated to housing the sensors at different levels. There were two cases per node which were mounted on an aluminium profile that was then attached to the ground support.
After some validation issues observed during the first year, some improvements were made to the mechanical enclosure to guarantee the water tightness of the plastic cases. There were no water intrusion problems or cable disconnections in the last round of field trials in spite of their transport and installation. The system comprised a modular design in which the electronics could be assembled and disassembled without major difficulties and with no connectivity issues.
The temperature and relative humidity sensors were also encapsulated in a final design made of PTFE caps which guaranteed the required air flow and waterproofness for a correct operation of the sensors. The antenna and the PV cell were located in the top part of the node to ensure, respectively, that minimum signal attenuation took place and the necessary sunlight reached the energy harvesting system. To ensure the waterproofing of the system, three piece systems with rubber joints were used in those parts where the water was in contact with the probe. All the nodes were painted in green to reduce the risk of theft and to hide the system from the birds that proved to be a threat to the system mechanical set-up.
Finally, the paddy nodes were modified to become repeater nodes where no sensor case was provided only a profile with a small communications and energy system was used.
The fifth objective was to design a flexible, easy to use, data-driven web-interface and back end software in order to process the retrieved data, as well as control instructions from operators. The web application was implemented using open-source web technology. The user interface system was based on MVC (Model-View-Controller) concept that isolates database with its processing (the Model) from the data manipulation (the Controller) and presentation of the data in web based user interface (the View). The model layer (where the database belonged to) defined the business logic of the system. The Controller system software architecture was a piece of code that called the Model to get some data that it passed to the View for rendering to the client. All requests were managed by front controllers. These front controllers delegated the real work to actions that were logically grouped into modules. The View was what the user interacted with. This layer was mainly made of PHP templates, which allowed a rapid development of views which were compatible with any viewing device (PC, phone or tablet) and any web browser. The view part of the Smart-Paddy software was developed in "open-source" technologies. The languages and structures which were selected were a MySQL database with PHP and AJAX (Asynchronous JavaScript and XML). As a data-interchange format inside the system JSON (JavaScript Object Notation) was used. It is a lightweight format, is easy for humans to read and write and for computers to parse and generate.
The data obtained through the sensor nodes was stored in the remote database and then the web-platform applied the tailored model developed within the project and assigned, for each node based on the environmental conditions. This was plotted graphically as pins in a map where the nodes were located providing in a simple and graphical interface a quick understanding of the rice blast infection risk in the area.
The sixth objective was devoted to weather-based rice blast forecasting models. After a thorough review of the existing rice blast forecasting systems and models, the final conclusion was that there is a limited number of them, which are nowadays operational. Thus, within 50 forecasting models, Yoshino model was decided to be integrated in the RICE-GUARD system. The choice was adjudged by the frequency of its appearances in the literature and in operation systems. Additionally, due to the simplified mode of approaching the most critical disease stage, the favourable conditions for the initial establishment of the pathogen on the rice plant (conidia penetration case). It should be stated that one of the most valuable forecasting outputs for the disease management is the prediction of the favourable conditions particularly in the beginning of the pathogenesis. This aspect is served very well by the functionality mode of the Yoshino model. Then, the model was coded on DEMETER’s server and migrated to IRIS one. The model’s predictions are based predominately on frequently recorded data of air temperature and leaf wetness, which were computed to daily wet risk hours, giving a final output of four levels risk range of favourable conditions, easily to be interpreted into risk colour geolocation maps. Besides the functionality mode, another very critical milestone to be succeeded in this part was the model’s thresholds that could sway the accuracy of the predictions. Therefore, after many field/laboratory tests, observations and assessments, the final threshold values were determined in order to modify the Yoshino model to accurate realise rice blast favourable conditions within the project’s European rice producing countries.
The seventh objective was the validation of the RICE-GUARD system in rice fields with different environmental characteristics. Consequently, after the development of the RICE-GUARD system in the terms of hardware, firmware and software, the integration of the modified Yoshino model was a very essential part of the project, since the rice blast favourable conditions forecasting is the final outcome of all project’s progress. Therefore, the predictions were tested and validated in-field at rice paddies of four countries Italy, Spain, Portugal and Greece. For this reason, rice blast data in the terms of initial disease appearance and progress were acquired by each local partner either by experienced personnel or by collaborating rice farmers. Then the procedure was carried out by comparing all the acquired data (disease and weather) with the model’s risk outputs. Thus, a more generic approach for Europe was used in the terms of thresholds strategy, while rice blast appearances were validated within the dates of high risk prediction outputs at each individual field location. Furthermore, the disease progress during the cultivation period obtained very good coincidences with the rest of the model forecasts until the end of the season. Ultimately, RICE-GUARD model predictions were compared, at four locations of Italy and Greece, with the collaborating project ERMES ones (GA: 606983). ERMES project has integrated the WARM model used in Joint Research Center in the project Monitoring Agricultural ResourceS (MARS, Ispra, Italy), which is used as generic forecasting model to map country loses due to rice blast and reports directly to the Commission. As a result, remarkably good coincidences in rice blast risk levels occurred at the four tested locations. Thus, RICE-GUARD in-field system occurred to be accurate and capable of predicting rice blast favourable conditions by two aspects: 1) by comparisons with the collected data derived from the project’s tested locations and 2) by the great coincidence of the predictions compared to the WARM rice blast model. Furthermore, it should be proposed for the future development of the project, to extend the data acquision (rice blast and canopy monitoring data) in order to intensify the model feed of the database. Moreover, more places should be included in the acquision, particularly to include as many as possible data collection sites and rice producing countries, which were not part of the RICE-GUARD project.
Finally, the eighth objective was to outline post project scaling-up rules and development work for full production. The results are described in deliverable D5.5 and include recommendations for the hardware, the model and the whole system. It is worth mentioning that discussions with farmers suggest that a prediction tool based on data remotely obtained by in-canopy data loggers and local weather forecasts could provide a large benefit with a minimal expense.
From the results above, the following are assumed to be novel enough yet not critical for the industrial exploitation of the RICE GUARD system:
1. Design of a solar energy harvester to measure solar irradiance.
2. Multiplexing a low-power solar Maximum Power Point Tracker for non-aligned solar panels.
3. A novel data logger with wireless wake-up and data downloading.
UPC proposed to publish a journal paper for each topic but one partner opposed to the third one because he assumed that the solution was original. In fact, only the application was original because the solution is based on components used in commercial remote control devices. Therefore, UPC will ask again for permission to publish it, and the other two.
DEMETER has submitted an 80-pages journal paper under the title: “Rice blast forecasting models and their practical value. A review”, to an open access journal “Phytopathologia Mediterranea” (2016 impact factor 1.32). The paper was based on the review of literature conducted during the Task 5.1. The manuscript was reviewed and currently it is at the second reviewing stage, while a revised version is expected to be resubmitted for the final decision from the journal.
The RICE-GUARD system serves the purpose it has been developed for. It is a smart system able to measure in-field conditions and provide decision support recommendations based on a tailored model to predict rice-blast appearance risk. Therefore, it provides an excellent tool to manage diseases in rice fields either for individual farmers or for farmer cooperatives or other associations. It can also be used as a research tool to study the effect of vegetation or crops in the intrinsic weather conditions and variations within the canopy.
Potential Impact:
There are numerous socio-economic impacts that will be derived from the results of this RICE-GUARD research project.
Central to the expected socio-economic impacts is the boosting of the competitiveness of rice farmers and companies along the rice value chain- by improving the crop yield and quality. Since the reduction of the intervention price for rice by 50% in 2004, the rice sector has had to readjust to become more competitive in front of the Asian market imports. Overall, the European rice sector faces a delicate situation and needs to benefit from any technology or system capable of enhancing their yield and therefore overall income.
Despite prices of rice falling back from the average prices in 2008 and 2009 they remained well above historical price levels, with the price in 2015/16 being between 100% and 110% above the values of 2002-04 but with a negative tendency from 2014. In 2016 the global trade of rice is expected to fall 3% with respect to the also reduced 2015 level and it is not likely to recover in 2017. Rice prices will also be lowered with respect to 2015. These figures suggest a considerable price volatility on global rice markets and reflect a strong dependence on climatic and weather related factors. This means that any system capable of predicting the natural variables of rice production –such as fungal infection, will be a very valued technology by the rice-growing sector. This need is enhanced by the fact that the trends in the EU remains constant in the fact that the consumption is growing faster than the domestic production, therefore imports are larger and include Indica or fragrant varieties. Thus, EU rice growers are in need to increase productivity and yield while they have to offer value added to their rice and this can be achieved through green rice and sustainable food products labelling. Meanwhile, the rest of the rice-growing continents need to ensure a constant activity in order to meet with the needs of the market and the world population. RICE-GUARD can become a valuable tool to fulfil all these needs across the global rice-growing community.
Taking into account the economic aspects and the benefits for the rice growing industry by achieving for example an improved fertilisation scheme or an optimised fungicide use, the RICE-GUARD technology is expected to have a significant impact on the rice growing industry, which is demanding of such improvements to cope with the new regulations in chemical phytoprotection products, reduce production costs and ensure food safety, and will therefore have a direct positive effect of the environment.
Moreover, the impact of RICE-GUARD will not only affect EU rice-growers, but it can have a global effect on the worldwide rice growing community. Production is geographically concentrated in Western and Eastern Asia with more than 90 % of world output. China and India, for example, account for more than 50% of the total world rice production. Brazil is the most important non-Asian producer, followed by the United States and the EU. If the RICE-GUARD system owners (4 SME associations and 3 other participants) manage to commercially exploit the system with a global patent, the sales could boost to all the producers across the rice-growing countries reaching a millionaire scale. This would imply a large benefit for the participating SME-AGs and SMEs while ensuring an optimised management of fungicides a safer food source in rice paddies worldwide.
The economic impact associated to RICE-GUARD will expand to several sectors across the value chain apart from the rice-producing end-users:
i) Producers of raw materials, components and equipment:
The electronics components and PCB producers will benefit from the demand of components for the RICE-GUARD system which will require the elements for the multiple parameter sensors, the energy management system and the RF transceiver system together with the solar cells for energy harvesting. Moreover, the PVC and PTFE producers and suppliers will also have to provide the necessary elements for the mechanical enclosure of the nodes.
ii) Manufacturing and distribution companies:
The large industrial manufacturing companies will have to produce and program the RICE-GUARD hardware and software based on the owners’ specifications and deliver an off-the-shelf package for the final retail or industrial agricultural companies to distribute the system to the final users. This will increase the overall activity in the value chain increasing their profit substantially.
iii) Agricultural machinery suppliers
The responsible companies for delivering the RICE-GUARD system to the final users will be those dealing directly with rice-growing associations, mainly supplying the agricultural tools and equipment needed for their operation. They could also be supplying the RICE-GUARD system as an innovative tool for the rice-growing communities and therefore increasing their share of sales and support and in return, their benefits. These companies will be the ultimate ambassadors and sales representatives of the RICE-GUARD system.
The increased productivity and growth generated by the more accurate prediction of rice-blast appearance will directly affect the job creation and rural employment, offering a social benefit to the RICE-GUARD system. In areas where rice is cultivated it is often the only source of income of the communities inhabiting those zones and therefore tools such as RICE-GUARD will be welcome by all agricultural practitioners and associations. Moreover, taking into account that rice provides 21 % of global human per capita energy and 15 % of per capita protein; that rice is the second largest produced cereal in the world and that world rice consumption increased, in the last 40 years, from 50 kg per capita to about 62 kg per capita (milled rice) (and this increase is goes up to 40% in Asian countries with more than 500 million people in hunger) we can assume that the introduction of tools such as RICE-GUARD offering an increased productivity will revert in an superior food security and production, contributing to securing the source of food for the growing world population.
In addition to the socio-economic benefits, the RICE-GUARD technology offers also environmental profits. Firstly, it will optimise fungicide application by suggesting when it is advisable to spray fields if the conditions for infection are taking place and, in some cases, lead to the reduction of fungicide use. Secondly, it can reduce the energy and soil impact from the fertilisation techniques. Since the fertilisation is directly related to the susceptibility of the plants, the application of N or P products can be decided on the basis (but not only) of the conditions of disease susceptibility. Thirdly, the availability of on-line data on the in-field conditions can save a lot of man-effort, and thus energy, in performing field trips for retrieving or measuring data. Finally, the concept of smart and sustainable agriculture will be enriched by the usage of RICE-GUARD. This will stimulate the consumption of “greener” products in the market and will have a positive effect on the environment, the economy and in food safety.
Finally, the RICE-GUARD sensor technology can be exported to other economic sectors such any other fungal or disease susceptible crops as well as any industrial sector requiring precise, on-site parameter monitoring. This could definitely become a major source of profits for the owners of the system while offering an improved productivity and yield to those sectors using the RICE-GUARD technology.
Main dissemination and exploitation activities
As previously discussed, the foreground generated by the RICE-GUARD project has a great potential and covers the whole supply and value chain while gives rise to potential exploitation opportunities for each member of the consortium. Each partner has expressed its intentions and wishes for the exploitation strategy. In addition, it allowed a large amount of dissemination activities to be carried out.
The current status of the protection and exploitation strategy for the RICE-GUARD foreground within the Consortium can be summarized as follows:
- The two industrial SMEs (OSV and LNL) will be responsible of commercializing and producing the system as described in the Joint Exploitation Agreement.
- The end-users SME-AGs are willing to have preferential access to the technology for using it in their rice paddies.
- An end-user SME (CAMARA) is willing to have access to the technology for their rice farmers.
All these roles and strategy are defined in the Joint Exploitation Agreement signed by all partners.
A really successful dissemination strategy has been carried out during the RICE-GUARD project maximising the exploitability of the material and demonstrating its potential to provide rice growers a tool for remotely controlling the status of their fields in terms of fungal susceptibility and thus increasing productivity and quality of rice.
Over the course of the project a total of 2534 interested people were directly contacted by the RICE-GUARD Consortium members in conferences, trade fairs, personal meetings and workshops. Moreover, a larger number were indirectly aware of the SMART-PADDY project via 6 press releases in the general and dedicated media, 3 specialized website publications, 1 radio interview and 3 articles pending publication in a scientific publication. The details can be found in the list annexed or in the Part A of the PUDF.
The project has attracted a large interest from the rice-growing and scientific community and has been in contact with other FP7 projects (ERMES project, Grant Agreement no. 606983 to exchange results and validate the produced model and create synergies among different technologies in agricultural production.
List of Websites:
www.riceguard.eu
Contact details:
Mr. Pau Puigdollers
IRIS Innovació i Recerca Industrial i Sostenible, Spain
ppuigdollers@iris.cat
The RICE-GUARD project developed an on-line, wireless sensor network in-field environmental monitoring system capable of measuring several environmental parameters (temperature, relative humidity, solar radiation, leaf wetness, among others) from inside the rice canopy. This system provides unprecedented data from different levels within the rice plants and is capable of substituting data logger nodes located in weather stations outside the plants (where the conditions are much different than inside the canopy) and single parameter measuring hand-held meters used when collecting accurate data inside the rice areas. This information could be used to model related factors affecting the crops such as the presence of pests, fungal diseases or other plant pathogens. In this regard, in combination with the hardware prototype developed, the RICE-GUARD system has also developed a tailored model for determining the risk levels of rice-blast (Pyricularia oryzae) incidence. One of the most important features of the system is its ability to provide remote readings of the environmental condition within the rice, as well as the level of risk associated to those, in a timely manner. Rice-growers have been able to read their local, in-field conditions using a tailored user interface which graphically represented the level of susceptibility to rice blast under which their rice fields were growing. This can result in a ultimate optimisation of the management of rice crops, ensuring a rational and efficient use of the commonly used, and potentially harmful, fungicides and chemical protective substances.
The RICE-GUARD system is composed by autonomous nodes containing 4 levels of temperature, relative humidity, leaf wetness and solar irradiance sensors, a communications module and an energy harvesting unit made of solar panels; an autonomous master node, equipped with all the sensors commonly found in weather stations plus a communication system receiving the readings of all nodes in the area and storing the information on a local database while uploading it to the cloud; and a user interface displaying the information coming from the field and applying the rice blast prognosis model.
RICE-GUARD was validated in 5 rice cultivation areas in Europe and proved to provide valuable information to all of them. It was installed in rice paddies under controlled and non-controlled conditions in order to prove the adaptability of the model. This allowed to effectively develop a model that was easily adaptable to all areas and which was demonstrated to predict the risk conditions in which the crops where infected. This was reinforced by the positive cross check of data with another European project, ERMES.
It was demonstrated that the RICE-GUARD system is able to provide much more accurate information than sensors located near fields but not inside them while allowing a real time, remote and user friendly monitoring of this information. This functionality was proven to work for managing rice blast incidence in rice crops and therefore increasing the yield, optimizing the physiological characteristics of rice and producing a safer food product.
The RICE-GUARD system that led to a route to market plan will need further validation, especially on the model testing, since creating robust models for reliable agricultural decision making takes usually between 5-6 years.
Project Context and Objectives:
Rice is the main crop in wet areas such as river deltas and is an essential tool in Europe in managing protected ecosystems. Irrigation water is a key factor in the production of rice and water quality has a major impact on crop yield as a result of tolerance of rice to factors such as dissolved salts. Rice is more water consuming than many other crops; in continuous flooding cultivation it takes about six times the water required by wheat. Due to increased water use in coastal areas, the sea intrudes the water table and rivers, and seawater floods nearby fields during storms in the Mediterranean area. The result is increased water salinity, which reduces yield in rice crops and increases soil salinity. Nowadays, water condition is for the most part assessed by visual inspection of the crops and, when excess water salinity is suspected, fields are irrigated by flooding them. In areas where water salinity is endemic, rice paddies are continuously irrigated with river water to reduce water salinity. This is a remedial solution that requires enormous volumes of water and considerable energy to pump water.
Water salinity can be accurately determined by measuring its electrical conductivity (EC). Measuring EC at the water inlet and outlet of each paddy field can help in monitoring the “washing” effect of irrigation. Moreover, measuring EC at points far from water inlets and outlets can help in assessing water salinity in a given paddy field and at different depths in drainage channels can help in managing water salinity in larger areas.
Based on this background the RICE-GUARD project built upon the firstly developed prototype system based on a master node (MN) and a series of paddy nodes (PN) and repeater nodes (RN) to validate the fabrication of the final RICE-GUARD system. On the other hand, the system was used to obtain the data needed for the rice blast prediction model refinement during the second rice season of the project. Two complete RICE-GUARD systems were installed and tested in Greece and Seville, while series of autonomous data-logging PN were installed in Portugal (2), Italy (1), Turkey (1) and Greece (3).
The main objective of RICE-GUARD was to obtain local information than can improve the accuracy and timeliness of existing rice blast forecasting models that currently rely on information sometimes obtained at more than 50 km from rice fields.
Firstly, the needs and specifications of the rice-growing industry were defined to drive the requirements of the different parts of the system. Initial requirements and specifications were analysed and after a series of consultations and after testing with a preliminary system, the final specifications were obtained. For the paddy nodes these included 4 measurement heights, with temperature, relative humidity and solar radiation so that at least 2 of them will be inside the rice canopy even in short varieties. The range of metrological specifications are 5 to 35 ºC for the temperature sensors, 70 to 98% for the relative humidity sensors and 0 to 1500W/m2 for the solar radiation ones. A leaf dew sensor was also deemed necessary due to the importance of the dew appearance which stays on the plant leaves. The nodes should be solar powered and with a communication system operating in the unlicensed ISM band at 433MHz. The master node should be able to work as a weather station all year long and should include temperature (-10 to 55 ºC), relative humidity (5 to 100%), solar radiation (1 to 1500 W/m2), wind speed (0 to 65 m/s), wind direction (0 to 360 º), rain precipitation (0 to 125 mm/h) and atmospheric pressure (660 to 1070hPa) sensors. It also should be prepared for mains power connection as well as solar powered and prepared for 3G or Wi-Fi connectivity. The user interface should show graphs on the mean readings in the field, graphs for one specific node and possibility to select periods of time to be graphically represented.
The sensors, their power supply system and the in-field user interfaces were developed following the designs and verification tests performed at laboratory scale. A novel remote wake-up system for the Bluetooth interface has been designed and successfully tested, achieving up to 22m communication range, significantly beyond the expected 10m. Similarly, very good results have been achieved in reducing the power consumption of the master node supply and sensor unit (SSU) and size (50%) compared to the preliminary Compact Weather Station system, despite the extra leaf wetness sensors.
An on-line wireless system for the measurement of salinity was designed and developed based on the requirements defined initially. This wireless sensor network (WSN) comprised independent nodes with Radio Frequency (RF) technology that communicated with a gateway device which gathered all data coming from the sensors and stored it locally and in a web server. It also included repeater nodes in order to cover a longer distance in large rice fields. The WSN was optimised in terms of distance of communications, message routing protocol and power consumption to meet the overall requirements of the system. Moreover, several autonomous data logger paddy nodes were designed and included temperature, relative humidity and leaf wetness sensors.
The newly developed sensors and WSN system were integrated into tailored mechanical enclosures which comprised the RICE-GUARD paddy and master nodes. Two new PCBs were designed (one for the communications control and one for the sensors and power supply system) following the research and tests performed previously. The best solution in terms of design simplicity were chosen in order to obtain an easy to assemble and disassemble device. A PVC encapsulation was designed for the nodes which offered the required water proofing capabilities. Similarly, a final PTFE encapsulation solution for the T and RH sensors was found. The system was built to allow laboratory and field testing. An accessible web interface was also developed to show the data obtained in the field in a graphical and understandable way as the front-end of the RICE-GUARD system.
Following a thorough literature review, the Yoshino model for rice blast appearance was decided to be adapted for local rice blast prediction among 50 analysed models. This model granted simplicity and customizability in order to make it widely adopted in all the project regions. The model comprised several parameters such as temperature range, time and wet hours to generate a level of alert directly related to the susceptibility of the plant to become affected by the fungus. After the selection and tailoring of the model, the RICE-GUARD system was tested during two rice seasons to obtain relevant data for enhancing the hardware while refining the model behaviour. Based on these results, recommendations for future improvements to obtain a fully commercial version of the system were outlined.
The process of validation of the RICE-GUARD system took place in 2 locations of the end-users SME-AG partners in the project and involved the testing of the system in rice paddies (Greece and Seville). The custom remote data loggers were tested in the relevant locations for the validation of the model such as rice paddies in Greece, north of Italy and Portugal. This testing at different sites of rice cultivation allowed to evaluate the adaptability of the model and to demonstrate the RICE-GUARD performance and features both within the consortium and beyond to a large number of industry stakeholders from the rice-growing sector.
The RICE-GUARD system proved to be an accurate and reliable tool for the management of rice paddies in rice-blast disease conditions and generally for improving the quality of rice. The close monitoring of the conditions for infection during the cultivation stage allows for a better regulation of the application of fungicides and thus optimises its use and reduces the potential negative environmental effect and at the same time improves food safety.
While developing the RICE-GUARD technology, concern was also given to the cost of selected materials and components to guarantee the future exploitation of the RICE-GUARD system by keeping it of sufficiently low cost to ensure its uptake by the rice growing sector. The developed system cost starts around the 5000€ per complete unit (including gateway, 3 paddy nodes, 1 repeater node and software package) which is in the range accepted by rice growers associations for such a device.
The overriding goal of this project was to ensure that the resulting pre-competitive RICE-GUARD prototype fulfils the set requirements to ensure its further development post-project into a fully industrial system that is taken to market, where its beneficial environmental and socio-economic impact will be felt at European level.
Project Results:
The two main results of the RICE GUARD project have been 1) the demonstration of the remarkable difference between the environmental conditions inside the canopy of inundated rice fields and those above the canopy, not to mention conditions in weather stations far away, and 2) the good prediction capability of the mathematical model developed. These results are relevant because all forecasting models described in the bibliography had been developed from weather data measured outside the canopy, often obtained from public weather stations some kilometres away from rice fields or derived from experiments in greenhouses. The results in RICE GUARD seriously question the scientific validity of studies not based on data measured inside the canopy. Since other fungal diseases in rice plants are also related to the conditions inside the canopy and the correlation between these and the external conditions is very poor in inundated fields, these two results are a leap forward in rice blast combatting and in precision agriculture at large.
These results have required several technological developments that can be applied to other crops and imply a change in paradigm for further developments in plant disease prevention and treatment. In order to describe the specific results achieved, the eight initial scientific and technological objectives of the project are summarized below:
1. To review the technical requirements of the members of the SME-AGs in order to define the system specifications for the RICE-GUARD prototype.
2. To develop weather sensors able to provide the information required by current rice blast forecasting methods that was not available at a local level.
3. To develop the telecommunication system to ensure the effective transmission of data from the RICE-GUARD sensors in-field to a remote data repository.
4. To design the mechanical enclosures for paddy nodes and the master node, to integrate the electronic subsystems, to assemble and test both types of nodes to verify the fulfilment of the electrical, mechanical and metrological specifications.
5. To design a flexible, easy to use, data-driven web-interface and back-end software in order to process the retrieved data, as well as control instructions from operators.
6. To assess weather-based rice blast forecasting models obtained from weather data in other regions in order to customize them by using local weather and additional parameters.
7. To fully test and validate the RICE-GUARD system in rice fields with different environmental characteristics.
8. To meticulously outline post project scaling-up rules and development work for full production.
The research to achieve of the first objective confirmed that no commercial forecasting system neither scientific works are based on real time local weather data obtained within the canopy as offered by a system such as RICE GUARD. The closer approach were measurements with thermo-hygrometer data loggers placed at or close to the border of the rice paddies and manually downloaded weekly, which had been used for some years by one of the SME-AG partners (FAS, Spain) but lacked a scientific validation, i.e. no peer-reviewed scientific publications. Another approach relied on weather data obtained by conventional weather stations inside the rice paddy with sensors above the canopy and some unspecified sensors somewhere within the canopy . The system was part of an extended network wherein data for some regions was extrapolated from measurements at nearby areas.
The research by questionnaires and interviews to farmers demonstrated that they were seriously worried about future regulations that could totally ban the most used fungicide (tricyclazole) and acknowledged the need for forecasting systems with reliable local predictions able to overcome the delay between infection and the appearance of obvious leaf lesions due to pyricularia oryzae. The lack of information about the range, accuracy and resolution of measuring equipment used to derive forecasting models, already detected in the literature review during project conception, as well as about actual environmental conditions within the canopy were confirmed by a more detailed literature review in Work Package 1. Therefore, for each of the three agreed testing sites a compact weather station (CWS) and a moveable field node were designed that together with a handheld node allowed the estimation of the measurement ranges of the environmental parameters involved in the models, their dependence on the position within the canopy, and eventually identify any other factors worth considering when defining the detailed system specifications. The main results of field measurement during the first rice season (2014) were:
1) The large inherent uncertainty of in-field measurands compared to that of the sensors available to measure them meant that there was no need for high-accuracy sensors. Uncertainty for in-field air temperature and relative humidity arises mostly from direct solar radiation and wind hence rice variety, as the height of rice plants ranges from 80 cm to about 120 cm. Commercial instruments having 0.1 °C and 1 %RH resolution seldom yield stable readings in open air, even if there is only light air, in spite of their verified stability inside a climatic chamber in the laboratory.
2) Air temperature and relative humidity inside the canopy are very different from those above it and cannot be inferred from measurements outside the canopy, e.g. those performed by a master node outside rice paddies.
3) Air temperature and relative humidity inside the canopy are determined by water temperature on the one hand and air and relative humidity above the canopy, as well as solar radiation and wind, on the other hand. Measurement at four heights above water level are enough to determine temperature and relative humidity profiles provided at least one measurement point is above the canopy. Measuring water temperature helps in that determination.
4) Air temperature and relative humidity above the canopy can be considered constant only for quite homogeneous areas. Otherwise, it has been verified that data from two public weather stations 10 km apart, one at sea level and the other one about 400 m above sea level can be very different whereas two stations less than 6 km away from each other but both at sea level yield closer readings the same days. Clouds yield fast temperature and relative humidity changes, so that partly cloudy sky results in heterogeneous conditions even in relative small areas that otherwise could be considered homogeneous. Wind speed undergoes large changes even for relatively close measurement points.
5) Since fungicide spraying decisions depend not only from past weather conditions but also from future weather predictions, particularly rain prediction, if no public weather station covers the rice farming area then weather sensors in the master node can be used for that prediction using appropriate meteorological models.
6) Sensors placed within 1 m from the border of a paddy field do not provide representative results of actual canopy conditions. Measurements must performed deeper in the field, which means remote control and data downloading are important to preserve genuine environmental conditions that could be modified by personally accessing the measurement points after installation.
7) Upon seeing the first-year field measurements, industrial partners showed a high interest on the capability of downloading data from in-field sensors even in the absence of Internet access. Consequently, this peer-to-peer communication capability was added to paddy nodes and the master node able to support Android and Windows devices. Both designs were based on off-the-shelf components and that for paddy nodes was particularly compact and low cost, hence offering remote operation capability to any data logger within a range of about 20 m.
The second objective was to develop weather sensors for air temperature (T) and relative humidity (RH), and solar radiation at different heights with respect to ground in paddy nodes placed inside rice fields, and also air temperature and relative humidity, atmospheric pressure, wind speed and direction, rain precipitation and CO2 concentration in master nodes outside the field but close to it. Leaf wetness was detected by calculating dew temperature from air temperature and relative humidity. The correlation between dew detection and leaf wetness detection using commercial sensors was excellent but the method based on T and RH could not detect leaf wetness duration. This raised the need for low-cost leaf wetness sensors.
The major design constraints for paddy node sensors were low cost and power consumption. Commercial integrated T and RH sensors with digital output are inexpensive, but no low-cost pyranometers for solar radiation are marketed. Consequently, it was decided to investigate the use of the very same solar cells used to harvest solar energy to measure the incoming energy flux. This required the solar panel to be split in two halves and measure the respective currents. By carefully orientating each panel, the direct and diffuse solar radiation were correctly measured and the cost of the electronics for power processing was reduced by multiplexing some circuits common to both panels. Field installation, however, was lengthy and needed some skill. Further, any movement of the node that changed the orientation or inclination of the solar cells led to the wrong results. Consequently, whereas the measurement of total solar radiation yields acceptable results, the distinction between direct and diffuse radiation works correctly only for fixed devices.
For the master node, power consumption was less restrictive but cost was still a constraint as usual in industry-oriented research. Therefore, building a weather station by assembling existing meteorological instruments would had been too expensive because of the redundancy of power supplies and communication interfaces of the several instruments required. The approach was to use different commercial sensors and design their electronic interfaces to build a sensor unit (SU) connected to a control and communication unit (CCU) in the same master node. The most cost-effective SU comprised a compact weather sensor (T, RH, barometric pressure, wind speed and direction, and solar irradiance), a CO2 concentration sensor, a rain gage and a leaf wetness sensor.
The third objective was to develop the telecommunication system able to effectively transmit data from the RICE-GUARD in-field sensors to a remote data repository. Initially, a MESH (Multipoint Enhanced Signal Handling) of nodes was expected to be used, but the two planned installations were configured in a linear way requiring a simpler network. The initially adopted solution in the end comprised Radio Frequency (RF) communication between the sensors and the Gateway in the unlicensed 433 MHz ISM band with a multi-hopping message relay protocol between paddy nodes. The selection of this transmission frequency was made according to the price and capabilities criteria. In the market there are several ISM transceivers available which are cheap and at the same they offer important features like high sensibility, low power consumption and programmable transmission power. This satisfied two important requirements of the RICE-GUARD system: it offered low-cost sensor nodes and an autonomous functioning over a long period of time. Moreover, the use of the ISM 433 MHz frequency allowed the implementation of a small-sized antenna that can be easily embedded in the sensor node mechanics.
In the second and third years, a modification to the initial protocol was made although the RF band and technology was maintained. This improvement was implemented to ensure that no data was lost in the multi-hop communication. The new protocol was based on Ad hoc On-Demand Distance Vector (AODV) routing. In this way, sent messages are acknowledged by the receiver and communication only takes place when requested by the master node, improving energy management. Moreover, through these system paddy nodes were able to act as repeater nodes.
The gateway located in the Master Node was the central device gathering all the data coming from the sensors and storing it locally while uploading all information to the cloud server. It was formed by an Alix 3d3 embedded PC system board with an RF transceiver and a USB modem. It used UMTS/GPRS or Wi-Fi protocol to communicate externally with the cloud server.
The fourth objective was to design the mechanical enclosure for paddy nodes and the master node and assemble and test the electronic subsystems before installing them in the field. Three separate designs were created for the Master Node, Paddy Node and Repeater Node.
The master node had all the electronics enclosed in an IP68 plastic box which provided the necessary water tightness for the different PCBs contained in it. The support was based on an off-the-shelf solution adapted for holding the electronics case and the sensors.
The body of the paddy nodes was based on common (non-pressure) PVC cases with IPX3 protection to reduce assembly costs. One case was devoted to the communications and control electronics while another, larger case, was only dedicated to housing the sensors at different levels. There were two cases per node which were mounted on an aluminium profile that was then attached to the ground support.
After some validation issues observed during the first year, some improvements were made to the mechanical enclosure to guarantee the water tightness of the plastic cases. There were no water intrusion problems or cable disconnections in the last round of field trials in spite of their transport and installation. The system comprised a modular design in which the electronics could be assembled and disassembled without major difficulties and with no connectivity issues.
The temperature and relative humidity sensors were also encapsulated in a final design made of PTFE caps which guaranteed the required air flow and waterproofness for a correct operation of the sensors. The antenna and the PV cell were located in the top part of the node to ensure, respectively, that minimum signal attenuation took place and the necessary sunlight reached the energy harvesting system. To ensure the waterproofing of the system, three piece systems with rubber joints were used in those parts where the water was in contact with the probe. All the nodes were painted in green to reduce the risk of theft and to hide the system from the birds that proved to be a threat to the system mechanical set-up.
Finally, the paddy nodes were modified to become repeater nodes where no sensor case was provided only a profile with a small communications and energy system was used.
The fifth objective was to design a flexible, easy to use, data-driven web-interface and back end software in order to process the retrieved data, as well as control instructions from operators. The web application was implemented using open-source web technology. The user interface system was based on MVC (Model-View-Controller) concept that isolates database with its processing (the Model) from the data manipulation (the Controller) and presentation of the data in web based user interface (the View). The model layer (where the database belonged to) defined the business logic of the system. The Controller system software architecture was a piece of code that called the Model to get some data that it passed to the View for rendering to the client. All requests were managed by front controllers. These front controllers delegated the real work to actions that were logically grouped into modules. The View was what the user interacted with. This layer was mainly made of PHP templates, which allowed a rapid development of views which were compatible with any viewing device (PC, phone or tablet) and any web browser. The view part of the Smart-Paddy software was developed in "open-source" technologies. The languages and structures which were selected were a MySQL database with PHP and AJAX (Asynchronous JavaScript and XML). As a data-interchange format inside the system JSON (JavaScript Object Notation) was used. It is a lightweight format, is easy for humans to read and write and for computers to parse and generate.
The data obtained through the sensor nodes was stored in the remote database and then the web-platform applied the tailored model developed within the project and assigned, for each node based on the environmental conditions. This was plotted graphically as pins in a map where the nodes were located providing in a simple and graphical interface a quick understanding of the rice blast infection risk in the area.
The sixth objective was devoted to weather-based rice blast forecasting models. After a thorough review of the existing rice blast forecasting systems and models, the final conclusion was that there is a limited number of them, which are nowadays operational. Thus, within 50 forecasting models, Yoshino model was decided to be integrated in the RICE-GUARD system. The choice was adjudged by the frequency of its appearances in the literature and in operation systems. Additionally, due to the simplified mode of approaching the most critical disease stage, the favourable conditions for the initial establishment of the pathogen on the rice plant (conidia penetration case). It should be stated that one of the most valuable forecasting outputs for the disease management is the prediction of the favourable conditions particularly in the beginning of the pathogenesis. This aspect is served very well by the functionality mode of the Yoshino model. Then, the model was coded on DEMETER’s server and migrated to IRIS one. The model’s predictions are based predominately on frequently recorded data of air temperature and leaf wetness, which were computed to daily wet risk hours, giving a final output of four levels risk range of favourable conditions, easily to be interpreted into risk colour geolocation maps. Besides the functionality mode, another very critical milestone to be succeeded in this part was the model’s thresholds that could sway the accuracy of the predictions. Therefore, after many field/laboratory tests, observations and assessments, the final threshold values were determined in order to modify the Yoshino model to accurate realise rice blast favourable conditions within the project’s European rice producing countries.
The seventh objective was the validation of the RICE-GUARD system in rice fields with different environmental characteristics. Consequently, after the development of the RICE-GUARD system in the terms of hardware, firmware and software, the integration of the modified Yoshino model was a very essential part of the project, since the rice blast favourable conditions forecasting is the final outcome of all project’s progress. Therefore, the predictions were tested and validated in-field at rice paddies of four countries Italy, Spain, Portugal and Greece. For this reason, rice blast data in the terms of initial disease appearance and progress were acquired by each local partner either by experienced personnel or by collaborating rice farmers. Then the procedure was carried out by comparing all the acquired data (disease and weather) with the model’s risk outputs. Thus, a more generic approach for Europe was used in the terms of thresholds strategy, while rice blast appearances were validated within the dates of high risk prediction outputs at each individual field location. Furthermore, the disease progress during the cultivation period obtained very good coincidences with the rest of the model forecasts until the end of the season. Ultimately, RICE-GUARD model predictions were compared, at four locations of Italy and Greece, with the collaborating project ERMES ones (GA: 606983). ERMES project has integrated the WARM model used in Joint Research Center in the project Monitoring Agricultural ResourceS (MARS, Ispra, Italy), which is used as generic forecasting model to map country loses due to rice blast and reports directly to the Commission. As a result, remarkably good coincidences in rice blast risk levels occurred at the four tested locations. Thus, RICE-GUARD in-field system occurred to be accurate and capable of predicting rice blast favourable conditions by two aspects: 1) by comparisons with the collected data derived from the project’s tested locations and 2) by the great coincidence of the predictions compared to the WARM rice blast model. Furthermore, it should be proposed for the future development of the project, to extend the data acquision (rice blast and canopy monitoring data) in order to intensify the model feed of the database. Moreover, more places should be included in the acquision, particularly to include as many as possible data collection sites and rice producing countries, which were not part of the RICE-GUARD project.
Finally, the eighth objective was to outline post project scaling-up rules and development work for full production. The results are described in deliverable D5.5 and include recommendations for the hardware, the model and the whole system. It is worth mentioning that discussions with farmers suggest that a prediction tool based on data remotely obtained by in-canopy data loggers and local weather forecasts could provide a large benefit with a minimal expense.
From the results above, the following are assumed to be novel enough yet not critical for the industrial exploitation of the RICE GUARD system:
1. Design of a solar energy harvester to measure solar irradiance.
2. Multiplexing a low-power solar Maximum Power Point Tracker for non-aligned solar panels.
3. A novel data logger with wireless wake-up and data downloading.
UPC proposed to publish a journal paper for each topic but one partner opposed to the third one because he assumed that the solution was original. In fact, only the application was original because the solution is based on components used in commercial remote control devices. Therefore, UPC will ask again for permission to publish it, and the other two.
DEMETER has submitted an 80-pages journal paper under the title: “Rice blast forecasting models and their practical value. A review”, to an open access journal “Phytopathologia Mediterranea” (2016 impact factor 1.32). The paper was based on the review of literature conducted during the Task 5.1. The manuscript was reviewed and currently it is at the second reviewing stage, while a revised version is expected to be resubmitted for the final decision from the journal.
The RICE-GUARD system serves the purpose it has been developed for. It is a smart system able to measure in-field conditions and provide decision support recommendations based on a tailored model to predict rice-blast appearance risk. Therefore, it provides an excellent tool to manage diseases in rice fields either for individual farmers or for farmer cooperatives or other associations. It can also be used as a research tool to study the effect of vegetation or crops in the intrinsic weather conditions and variations within the canopy.
Potential Impact:
There are numerous socio-economic impacts that will be derived from the results of this RICE-GUARD research project.
Central to the expected socio-economic impacts is the boosting of the competitiveness of rice farmers and companies along the rice value chain- by improving the crop yield and quality. Since the reduction of the intervention price for rice by 50% in 2004, the rice sector has had to readjust to become more competitive in front of the Asian market imports. Overall, the European rice sector faces a delicate situation and needs to benefit from any technology or system capable of enhancing their yield and therefore overall income.
Despite prices of rice falling back from the average prices in 2008 and 2009 they remained well above historical price levels, with the price in 2015/16 being between 100% and 110% above the values of 2002-04 but with a negative tendency from 2014. In 2016 the global trade of rice is expected to fall 3% with respect to the also reduced 2015 level and it is not likely to recover in 2017. Rice prices will also be lowered with respect to 2015. These figures suggest a considerable price volatility on global rice markets and reflect a strong dependence on climatic and weather related factors. This means that any system capable of predicting the natural variables of rice production –such as fungal infection, will be a very valued technology by the rice-growing sector. This need is enhanced by the fact that the trends in the EU remains constant in the fact that the consumption is growing faster than the domestic production, therefore imports are larger and include Indica or fragrant varieties. Thus, EU rice growers are in need to increase productivity and yield while they have to offer value added to their rice and this can be achieved through green rice and sustainable food products labelling. Meanwhile, the rest of the rice-growing continents need to ensure a constant activity in order to meet with the needs of the market and the world population. RICE-GUARD can become a valuable tool to fulfil all these needs across the global rice-growing community.
Taking into account the economic aspects and the benefits for the rice growing industry by achieving for example an improved fertilisation scheme or an optimised fungicide use, the RICE-GUARD technology is expected to have a significant impact on the rice growing industry, which is demanding of such improvements to cope with the new regulations in chemical phytoprotection products, reduce production costs and ensure food safety, and will therefore have a direct positive effect of the environment.
Moreover, the impact of RICE-GUARD will not only affect EU rice-growers, but it can have a global effect on the worldwide rice growing community. Production is geographically concentrated in Western and Eastern Asia with more than 90 % of world output. China and India, for example, account for more than 50% of the total world rice production. Brazil is the most important non-Asian producer, followed by the United States and the EU. If the RICE-GUARD system owners (4 SME associations and 3 other participants) manage to commercially exploit the system with a global patent, the sales could boost to all the producers across the rice-growing countries reaching a millionaire scale. This would imply a large benefit for the participating SME-AGs and SMEs while ensuring an optimised management of fungicides a safer food source in rice paddies worldwide.
The economic impact associated to RICE-GUARD will expand to several sectors across the value chain apart from the rice-producing end-users:
i) Producers of raw materials, components and equipment:
The electronics components and PCB producers will benefit from the demand of components for the RICE-GUARD system which will require the elements for the multiple parameter sensors, the energy management system and the RF transceiver system together with the solar cells for energy harvesting. Moreover, the PVC and PTFE producers and suppliers will also have to provide the necessary elements for the mechanical enclosure of the nodes.
ii) Manufacturing and distribution companies:
The large industrial manufacturing companies will have to produce and program the RICE-GUARD hardware and software based on the owners’ specifications and deliver an off-the-shelf package for the final retail or industrial agricultural companies to distribute the system to the final users. This will increase the overall activity in the value chain increasing their profit substantially.
iii) Agricultural machinery suppliers
The responsible companies for delivering the RICE-GUARD system to the final users will be those dealing directly with rice-growing associations, mainly supplying the agricultural tools and equipment needed for their operation. They could also be supplying the RICE-GUARD system as an innovative tool for the rice-growing communities and therefore increasing their share of sales and support and in return, their benefits. These companies will be the ultimate ambassadors and sales representatives of the RICE-GUARD system.
The increased productivity and growth generated by the more accurate prediction of rice-blast appearance will directly affect the job creation and rural employment, offering a social benefit to the RICE-GUARD system. In areas where rice is cultivated it is often the only source of income of the communities inhabiting those zones and therefore tools such as RICE-GUARD will be welcome by all agricultural practitioners and associations. Moreover, taking into account that rice provides 21 % of global human per capita energy and 15 % of per capita protein; that rice is the second largest produced cereal in the world and that world rice consumption increased, in the last 40 years, from 50 kg per capita to about 62 kg per capita (milled rice) (and this increase is goes up to 40% in Asian countries with more than 500 million people in hunger) we can assume that the introduction of tools such as RICE-GUARD offering an increased productivity will revert in an superior food security and production, contributing to securing the source of food for the growing world population.
In addition to the socio-economic benefits, the RICE-GUARD technology offers also environmental profits. Firstly, it will optimise fungicide application by suggesting when it is advisable to spray fields if the conditions for infection are taking place and, in some cases, lead to the reduction of fungicide use. Secondly, it can reduce the energy and soil impact from the fertilisation techniques. Since the fertilisation is directly related to the susceptibility of the plants, the application of N or P products can be decided on the basis (but not only) of the conditions of disease susceptibility. Thirdly, the availability of on-line data on the in-field conditions can save a lot of man-effort, and thus energy, in performing field trips for retrieving or measuring data. Finally, the concept of smart and sustainable agriculture will be enriched by the usage of RICE-GUARD. This will stimulate the consumption of “greener” products in the market and will have a positive effect on the environment, the economy and in food safety.
Finally, the RICE-GUARD sensor technology can be exported to other economic sectors such any other fungal or disease susceptible crops as well as any industrial sector requiring precise, on-site parameter monitoring. This could definitely become a major source of profits for the owners of the system while offering an improved productivity and yield to those sectors using the RICE-GUARD technology.
Main dissemination and exploitation activities
As previously discussed, the foreground generated by the RICE-GUARD project has a great potential and covers the whole supply and value chain while gives rise to potential exploitation opportunities for each member of the consortium. Each partner has expressed its intentions and wishes for the exploitation strategy. In addition, it allowed a large amount of dissemination activities to be carried out.
The current status of the protection and exploitation strategy for the RICE-GUARD foreground within the Consortium can be summarized as follows:
- The two industrial SMEs (OSV and LNL) will be responsible of commercializing and producing the system as described in the Joint Exploitation Agreement.
- The end-users SME-AGs are willing to have preferential access to the technology for using it in their rice paddies.
- An end-user SME (CAMARA) is willing to have access to the technology for their rice farmers.
All these roles and strategy are defined in the Joint Exploitation Agreement signed by all partners.
A really successful dissemination strategy has been carried out during the RICE-GUARD project maximising the exploitability of the material and demonstrating its potential to provide rice growers a tool for remotely controlling the status of their fields in terms of fungal susceptibility and thus increasing productivity and quality of rice.
Over the course of the project a total of 2534 interested people were directly contacted by the RICE-GUARD Consortium members in conferences, trade fairs, personal meetings and workshops. Moreover, a larger number were indirectly aware of the SMART-PADDY project via 6 press releases in the general and dedicated media, 3 specialized website publications, 1 radio interview and 3 articles pending publication in a scientific publication. The details can be found in the list annexed or in the Part A of the PUDF.
The project has attracted a large interest from the rice-growing and scientific community and has been in contact with other FP7 projects (ERMES project, Grant Agreement no. 606983 to exchange results and validate the produced model and create synergies among different technologies in agricultural production.
List of Websites:
www.riceguard.eu
Contact details:
Mr. Pau Puigdollers
IRIS Innovació i Recerca Industrial i Sostenible, Spain
ppuigdollers@iris.cat