Month: December 2020

We have the pleasure to release the second SWAMP Newsletter, to spread the word about the latest developments. The SWAMP project has been generating theoretical and mainly practical down-to-earth knowledge about applying IoT technologies, managing large amounts of data and using artificial intelligence algorithms and optimization techniques to create sensible water need estimates and water distribution schedules.

The SWAMP Project has officially ended on October 2020, but this is not a farewell message, since the project has been extended in Brazil up to April 2021 and members will continue to publicize results. Also, this webpage will be up and running for many years from now.

In the SWAMP Newsletter #2 we expect to convey valuable information to different interested stakeholders regarding the latest developments in the pilots, whose local efforts have been extremely impacted by the 2020 COVID pandemic. Also, this issue summarizes the results of 11 papers recently published by the SWAMP members.

If you are willing to obtain additional information regarding the SWAMP Project’s findings, please do not hesitate to contact us by any means, including the official project email at contact@swamp-project.org.

 

Sincerely,

 

Juha-Pekka Soininen & Carlos Kamienski

 

 

The objective of the SWAMP Italian pilot was the development and test IoT solutions for precise water-need estimation and optimal irrigation management to improve irrigation efficiency, reducing water usage and energy costs. The peculiarity of the Italian pilot lies on the presence of a Water District Manager (WDM), that is a public or private entity in charge of the management of a complex canal network devoted at allocating the water to several farmers (e.g., Consorzio di Bonifica dell’Emilia Centrale in the case of this pilot). In such context (common in many parts of Italy and other countries), a proper definition of the irrigation scheduling, together with the use of a water balance model for estimating crops water requirements, represent two pillars to try to reduce water usage in agriculture. IoT solutions developed during the project can serve as the framework collecting data from sensors in the fields, from infrastructures or external services (e.g., weather forecast, drone surveys) and providing tools and indications to farmers and WDMs.

These solutions have been implemented in the Italian pilot (892 ha, 320 ha of which are irrigated), which is located in the Po valley, near-by the city of Reggio-Emilia. Although COVID-19 pandemic significantly affected the project plans limiting real trials, SWAMP solutions proven their potential in water saving.

The precise water-need estimation is pursued within the SWAMP platform implementing the CRITERIA model, developed by ARPAE, thanks to which the amount of water saved in a pear orchard is estimated to be on average equal to 80%, without having any yield loss. In particular, the following figure shows precipitations and total irrigation water volume in the two irrigation seasons (2019 and 2020, panel a) and b), respectively), where orange and blue areas represent the cumulative irrigation water requirement simulated by CRITERIA and used by the farmer, respectively. Although not proven by real tests carried out in the field, these results shade some light on the potential of such solutions.

The optimization of the water distribution among multiple farmers and crops served by a common irrigation network is carried out adopting a mixed integer linear programming solution (MILP). MILP collects all water requests and provides gatekeeper with the optimal irrigation scheduling minimizing water loss by infiltration, delays on water delivery and field operations. The impact of the optimization algorithm has been evaluated referring to a series of real irrigations managed by the WDM during a 15-day period of 2020 season. The efficiency improvement results in a 32% reduction of the overall inflow volume required by the district. Grey curve in the following figure represents the irrigation request from farmers, while the blue one represents the volume used to satisfy such irrigation needs as recorded at the district inlet. Benefit associated to the adoption of the MILP algorithm is evident looking at the orange curve, which shows the volume that would have been required by the district, during the same period, in case of adopting the optimal scheduling.

Not withstanding the difficulties that characterized the final phases of the project due to COVID-19, the Italian pilot fully succeeded in proving the potential of IoT-based solutions to sustain both optimal water-need estimation and scheduling optimization approaches.

Dissemination Activities

 

Dissemination for SWAMP means the spreading of technical and scientific knowledge generated within a project, is essential for project outcome take-up, which, in turn is a measure of project impact. Dissemination activities focus on the publication of journal and conference papers, presentation of posters, organization of workshops, meeting and community events, and participation in scientific conferences.

The SWAMP members were very prolific in terms of actively participating in dissemination activities during the three years of the project, as highlighted by the table below. Internal dissemination activities include workshops, meetings, visits, and researcher exchange.

Dissemination	1st Year	2nd Year	3rd Year	Total
Journal Paper	1	4	3	8
Conference Paper	6	11	15	32
Poster	4	9	1	14
Workshop Organization	–	2	1	3
Internal Dissemination	6	6	2	14
Cluster/Community	7	3	1	11

 

Particularly, two papers published by SWAMP members received special attention from the scientific community. The paper “Smart Water Management Platform: IoT-Based Precision Irrigation for Agriculture” published by the Sensors Journal, caused a high impact in the scientific community since it was published in January 2019. As of December 8^th 2020, thisimpact can be measured by:

• 90 citations on Google Scholar
• 34 citations on the Web of Science
• 6737 abstract views and 8948 full-text views on the Sensors website
• Most viewed paper on the Sensors website in 2019 with 3189 views

 

Also, the paper “Water spray detection for smart irrigation systems with Mask R-CNN and UAV footage”, published in the IEEE MetroAgriFor 2020 in early November 2020 received the conference best paper award.

 

 

Communication Activities

 

Communication for SWAMP targets the internal activation and motivations of consortium partners both in Europe and Brazil, increasing the awareness of project activities, and for interacting with external interested and relevant organisations and persons. Communication activities cover the communication needs of the project towards partners and their direct networks, lead users, research partners, students, related projects, regional and national governments, policy makers, standardisation bodies and society, and the general public.

Given the subject of the project – IoT and agriculture – and its importance in sustainable food security for the world population, SWAMP has been receiving significant attention from the expected audience and from the media. Deliverables D6.3, D6.8, and D6.4 report the key communication achievements during the three years of the project. Nevertheless, some achievements are highlighted in the table below.

• The SWAMP members were actively involved in communication activities of different formats, as well as the project received significant coverage from the media.
• The website became the key source of information about the project, receiving a high number of unique visitors during the three years.
• Twitter was also useful for communicating project activities and achievements, receiving dozens of thousands of impressions.

 

Communication	1st Year	2nd Year	3rd Year	Total
Communication Activities	33	19	21	73
Website Visitors	1,185	888	1,055	3,128
Twitter Impressions	29,762	34,446	20,296	84,504

 

 

Exploitation Activities

 

Exploitation for SWAMP aims at guaranteeing that significant project results survive after the end of the project. Thus, the idea of this task is driving the consortium members to achieve the goals established in the beginning of the project and to account for new needs, possibilities and opportunities. The forecasted exploitation possibilities include a) using project results in further research activities, which are not covered by the project itself; b) developing and providing a product, process or service, which have a clear focus on the market; c) using project results in standardization activities and policy-making or advocacy actions.

The SWAMP consortium integrates different profiles of stakeholders in the area of IoT for agriculture, including farmers (Intercrop, CBEC pilot farms, Guaspari Winery, Rio das Pedras Farm (MATOPIBA), a water distributor (CBEC), a drone manufacturer (Quaternium), a system integrator (LeverTech), two research and technology transfer institutions (VTT and EMBRAPA), and four universities as scientific partners (UNIBO, UFABC, UFPE, FEI). This mix is considered a leverage for the exploitation of the results.

 

Important landmarks related to the exploitation of SWAMP results are:

• Internal Workshop on Innovation and Exploitation (Deliverable D6.7): held as a series of three online meeting during September and October 2018, collected the initial perspectives of all partners on SWAMP innovation and exploitation;
• Exploitation Plan (Deliverable D6.6): built upon the Internal Workshop on Innovation and Exploitation, this deliverable provides an overview and purpose of exploitation within the SWAMP project, identifying the key target audiences and highlighting the innovation potential;
• Final SWAMP Exploitation Workshop: occurred on October 2020, revealed that the views on exploiting SWAMP outcomes evolved significantly and some of them have been already materialized, as the creation of a startup in Italy and a variety of project proposals for extending the project achievements;
• Exploitation Activity Report (Deliverable D6.5): based on the final SWAMP Exploitation Workshop, reports the final views of SWAMP partners on the exploitation of project results, as well as exploitation activities that have already been carried out by the project members;
• SWAMP Interest Group (SIG): officially introduced with the Exploitation Plan and after some unsuccessful attempts to aggregate interested parties by invitation, an open call was issued that resulted in the current 34 members. A first newsletter was issued targeted to the general public, but specially to the SIG members
• Startup: VAIMEE is a startup founded in 2020 by a member of the SWAMP UNIBO team, aiming at providing B2B services in several domains (including farming), based on SWAMP technology.
• Projects: SWAMP spawned two research project proposals that are currently under review: one involving Brazil and Italy and the other involving Brazilian partners to complete de MATOPIBA pilot.
• Legacy: SWAMP partners believe the project left a significant legacy in terms of exploiting findings, understanding, insights, approaches, algorithms, and mostly lessons learned. Since the SWAMP partners engaged in many different dissemination activities, the knowledge generated within the project will survive the project official end and survive in the future

 

 

SWAMP Cartagena pilot focused on optimising irrigation water consumption when growing baby-leaf spinach in semi-arid area close to Cartagena, Spain. The idea was to measure the soil moisture data from three depths of the field, i.e. 5cm, 10cm, and 15cm, to feed the data to our SWAMP smart water management platform, create soil moisture forecasts and irrigation plans, and to irrigate the soil with IoT based automated sprinkler irrigation system.

As we all know, the COVID-16 pandemic with its almost global lockdowns destroyed these plans. we were not able to access the fields, and plans had to changed.

We had collected the soil moisture data and weather data, and we decided to simulate the soil moistures with various irrigation scenarios.  The soil moisture forecast model was created based on collected soil behaviour data. The model predicted how water flows through different soil layers from surface up to 50cm deep. As we had the weather data collected from pilot field from the February 2020, we simulated that time period with several irrigation scenarios. In the scenarios were varied in the irrigation parameters and irrigated the crop trying to maintain the root zone within target soil moisture range that was known to be good for baby-leaf spinach. Then we calculated how much water was needed.

The figure above illustrates the soil moisture development of all 10 5cm layers in the simulation case when initial irrigation to the dry soil was 80mm at February 1^st, and the minimum time between irrigations was 12 hours after the seeding of spinach at February 6^th. Irrigations were stopped at February 26^th when crop was harvested.

We simulated about 40 such scenarios and compared the results to the real farming data at the same time period. In the real farming time period, the soil moisture data and irrigation recommendations were not used. Based on the results, we can say that 25% of water saving is possible with very little risks. Even then we lose water to the deeper layers of the soil, but it cannot be avoided by changing irrigation strategy alone.

IoT and soil moisture measurement improves the knowledge of crops’ water needed and helps in saving the water that is a scarce resource in Cartagena and many other parts in this planet. Even with the simple automation, the sustainability of food production can be improved.

In February 2020, early before the Italian lockdown, the VAIMEE start-up has been founded. VAIMEE is going to become an official spin-off of the University of Bologna by the end of December 2020. VAIMEE aims to exploit SEPA (SPARQL Event Processing Architecture), an enabling technology of the SWAMP platform. SEPA development started in 2016 and the SWAMP project represented the ideal context for this solution to evolve, to be validated and to prove its information interoperability support.

“We provide B2B solutions for the development of interoperable services and applications on top of an open software solution driven by Semantic Web technologies and Linked Data standards.” (picture courtesy of VAIMEE srl)

A group of SWAMP researchers received the best paper award in the MetroAgriFor conference, held online from 4 to 6 of November 2020. MetroAgriFor aims to put together researchers working with measurement and data processing techniques for Agriculture, Forestry, and Food. The paper entitled “Water spray detection for smart irrigation systems with Mask R-CNN and UAV footage,” by Caio Albuquerque, Sergio Polimante, André Torre-Neto, and Ronaldo Prati uses a MASK R-CNN neural network for detecting the water spray from irrigation systems recorded using an Unnamed Aerial Vehicle (UAV), or drone, as illustrated in the Figure. In the future, farmers can use the system to inspect irrigation systems to detect failures, potentially reducing time and cost in system maintenance and verify the correct deployment of the irrigation plan.

The SWAMP project published a paper in the IEEE Global Humanitarian Technology Conference (GHTC 2020), reporting experience with fuzzy logic to control the automated irrigation of crops using soil moisture, air humidity, and air temperature sensor data. The fuzzy system was able to reduce about 20% of the water used for irrigation. For the second year, SWAMP dissemination activities reach this important IEEE conference focused on technology for sustainability and other humanitarian concerns, following the example of GHTC 2019.

This paper presents a fuzzy inference system based irrigation control system to determine water irrigation volume for a sweet pepper crop. The system is composed of a Mandani fuzzy control algorithm that receives data from soil moisture, air humidity and air temperature sensors from the crop field. Also, the rain forecast is gathered from a personal weather station located near to the crop field. Data gathered are used by the control algorithm to ensure that the soil moisture at the root zone is readily available to the plant.

The first results show that after 62 days from seeding, the system can reduced irrigation by about 20% through controlling soil moisture levels at the plant root zone, saving water content on the irrigation process.

The SWAMP project published a paper in the 2020 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor 2020) reporting a performance analysis of the LoRaWAN LPWAN technology for smart agriculture. Understanding the tradeoffs of LoRaWAN for IoT-based Smart Irrigation proposes of a two-step methodology to evaluate the performance of LoRaWAN based on simulation for understanding the behavior of the air interface and measurement for understanding the behavior of the IoT Platform. This paper provides insights to understand the tradeoffs of LoRaWAN deployments that may be useful to anyone considering starting a massive deployment of IoT smart applications, particularly in agriculture.

The picture below shows the scenario represented in the experiments of Step 1 and 2, namely the farm hosting the MATOPIBA pilot of the SWAMP project, with four circular parcels irrigated by a center pivot where corn, soybeans, and cotton crops take turns all over the year. Each parcel was evaluated independently of the others to avoid cross interferences among different distances and understand the effect of distance in LoRaWAN performance

We evaluate an emulated scenario with up to 10,000 sensors in conditions very similar to those found in field deployments. The LoRa air interface is captured by simulation using ns-3, whose results are fed into a testbed that emulates sensors transmitting LoRaWAN packets and using the typical software components of any real deployment. Our evaluation used a synthetic sensor workload and a simulated LoRa air interface, but from the LoRaWAN gateway on, the IoT Platform is similar to those used in real deployments.

Delay results for Parcel 4, varying SF and sensor density, reveal a general low growth rate when the number of sensors increases, but a high growth rate as the SF increases. On the other hand, packet loss skyrockets as the density increases, becoming intolerable from 1,000 on, as most packets are dropped.

The SWAMP project published a paper in the 2020 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor 2020) reporting an experience aimed at evaluating the Sensitivity of the agro-hydrological model CRITERIA-1D to the Leaf Area Index parameter for precision irrigation management. The CRITERIA-1D model is currently running into the SWAMP platform for estimating crop water requirements in the Italian pilot. The leaf area index is an input data of the model and is used to characterize the plant status and to represent its developing stages. To assess the importance of the LAI parameter on the estimation of the plant water needs, the model was set-up to mimic the agro-environmental conditions of the selected pear orchard. Two scenarios were defined: in the first reference scenario literature LAI values were used (S_ref); in the second scenario measured LAI values (S_LAI), collected with the AccuPAR LP80 ceptometer, were replaced to reproduce the real filed conditions. During the summer 2019, four sampling campaigns were conducted in the pear orchard and LAI measurements were collected in 16 points in a regular grid; for every point six measurements were taken around the plant.

Results of the simulations with the model are presented in the two figures below. As it is shown, for the modified scenario (S_LAI) LAI values were lower than those of the reference scenario (S_ref). This is accompanied by a decrease of the plant’s potential transpiration (left graph) resulting in less water demand for the modified scenario (S_LAI).

Looking at the irrigation water requirements, a lower number of irrigation events compared to the reference scenario have been quantified. Consequently, a lower total irrigation water need is also estimated, as it is possible to see in the right figure.

As general conclusion, the results confirmed the sensitivity of water need estimation CRITERIA-1D model output to the LAI values. Following this analysis, the model into the SWAMP platform has been updated using LAI ground-based measurements, to better represent the actual field condition as this parameter affects the management of irrigation water and its consequent saving.