Following the AgriDataValue online workshop conducted in March 2024 on the Digital Transformation in Agriculture, we have submitted a conference proceeding to the AgEng2024 conference held in Athens, Greece, from July 1-4, 2024. AgriDataValue has been acknowledged in the proceeding titled “Data-driven Solutions for Farmer Empowerment in Smart Agriculture: Challenges and Opportunities,” which is available at the following link: AgEng2024 Proceedings.
An online discussion was held on Thursday, 4h of July 2024, between the Horizon Europe projects AgriDataValue, ScaleAgData and CrackSense to explore new avenues for collaboration. The purpose of the meeting was to analyse opportunities for future cooperation and to examine potential synergies in dissemination and communication efforts. During the meeting, participants showcased their projects and discussed recent advancements. Among the participants of the meeting there were also the representatives of the National Paying Agency (Lithuania), which is one of the 30 consortium partners of the AgriDataValue project under the coordination of Synelixis Solutions S.A., Greece.
The connecting link of three projects, AgriDataValue, CrackSense and ScaleAgData, is that all of them are being implemented under the same Horizon Europe programme call: HORIZON-CL6-2022-GOVERNANCE-01 (Innovative governance, environmental observations and digital solutions in support of the Green Deal), which means that the above mentioned projects are targeting similar goals and objectives. Therefore joining efforts and sharing data as well as results of respective activities would be an added value for all three projects.
The main idea of the Horizon Europe project AgriDataValue is to establish itself as the “Game Changer” in Smart Farming digital transformation and agri-environmental monitoring, and strengthen the smart-farming capacities, competitiveness and fair income by introducing an innovative, open source, intelligent, multi-technology and fully distributed Agri-Environment Data Space (ADS). More info about the project you can find here: AgriDataValue
The main idea of the Horizon Europe project ScaleAgData is to develop the data technology (data streaming, data analytics, AI) needed to scale data collected at the farm level to regional datasets built for agri-environmental monitoring and the management of agricultural production. More info about the project you can find here: ScaleAgData
The main idea of the Horizon Europe project CrackSense: to ensure high throughput real-time monitoring and implementing in practice the prediction of fruit cracking by utilising and upscaling sensing and digital data technologies. More information about the project you can find here: https://cracksense.eu/
The collective goal of the initiative is to boost the visibility of data-driven agriculture and evaluate the potential benefits of data sharing. The representatives of the above three projects discussed options of mutual communication and agreed upon further joint actions.
During the 8th Regional Info Day on “The Adaptation of the Peloponnese Region to Climate Change” on June 20th in Kalamata, George Kokkinos, President of NILEAS Producers’ Group, presented the AgriDataValue project, which aims to revolutionize the agricultural sector through advanced technology at the European level. He also delivered a speech titled “Olive Orchards and Climate Crisis: Risks, Challenges, and Opportunities.” He highlighted the need for a comprehensive approach to ensure the sustainability of olive cultivation in the context of climate change. Mr Kokkinos concluded by emphasizing the lack of a coherent policy and plan for the agri-food sector, warning of the potential threat to agriculture due to the escalating effects of climate change. More than 150 people participated in the event, including policymakers, representatives of local authorities, cultivators, advisors, and scientific community members.
The second Community of Practices (CoP) was organized on June 12, 2024, by the Greek team of SUPPORT, Smart Farming Technology Group – Agricultural University of Athens, and NILEAS Producers’ Group. Vicky Inglezou, project manager of NILEAS’ Producers Group had the opportunity to present the AgriDataValue project. 17 representatives from the olive sector, i.e., farmers, advisors, and representatives of science, participated to discuss the IPM future scenarios of olive farming through an interactive co-creation workshop.
Panta Rhei Conference is a biannual event, during which the representatives of the EU CAP paying agencies share experience, good practices and discuss rising challenges with regard to computerisation, digitalisation, organisational issues and technical implementation of the measures, taken in the process of CAP implementation. The conference is organised twice a year: in spring and autumn. During the Panta Rhei conference of the EU CAP paying agencies in Antwerp, Belgium, 22-24 May 2024, the importance of the Horizon Europe project AgriDataValue for the future development of smart farming was highlighted in the presentation “What’s next after AMS?“ delivered by the NPA representative. NPA in its daily activities is promoting digitalisation and automation of agricultural practices. This strategy is particularly focused on the needs of rural areas. The solutions developed by the ADV project will support farmers in taking advantage of digital technologies and novel services, at the same time contributing to the climate change mitigation and adaptation. The introduction of innovations will increase the efficiency and competitiveness of farmers’ operations, will reduce GHG emission and thus contribute to achieving the environmental protection goals, specified in the EU strategy “Green Deal”, EU Climate Action and other EU strategic documents. The presentation received a positive feedback from the audience.
AgriDataValue partner, University of Lodz hosted the 30th Annual Colloquium of the Commission on Sustainable Rural Systems, International Geographical Union. This scientific event took place between 5th and 9th June 2023 in Lodz, Poland. After two days of paper and poster sessions focused on “Clases of Knowledge: Green Deal Concepts and Challenges for Sustainable Rural Areas”, over 50 conference participants representing 18 countries (Australia, Austria, Belgium, Brazil, China, Croatia, Czech Republic, Ecuador, Germany, India, Italy, Japan, Poland, Portugal, Romania, Spain, USA) went for a field session where specific challenges and solutions in sustainable development of rural areas in Central Poland were presented. During the trip we also visited one of the AgriDataValue Pilots – “The Wind Orchard” where AgriDataValue project was presented.
Given the urgency of climate change, the integration of climate projections and indicators has become imperative for sustainable agricultural planning. Europe, with its diverse climatic regions and intricate agricultural landscape, is a representative study area for this purpose. Accordingly, this study presents a methodology for downscaling and representing the most relevant indicators and projections for agricultural planning across Europe. Based on a detailed literature review, several indicators were retained ranging from the impact of temperature fluctuations on crop yields to anticipating shifts in precipitation patterns affecting water management strategies, and beyond for tapping into the intersection of climate science and agricultural resilience. This aspect is one of the pillars of the AgriDataValue project and ambition. Accordingly, using the CMIP6 model outputs, projections of the following climate indicators for the years 2030-2050-2070 under the SSP2-4.5, and SSP5-8.5 scenarios will be performed:
The Coupled Model Intercomparison Project Phase 6 (CMIP6) is a collaborative international effort involving climate modeling groups worldwide. CMIP6 aims to improve our understanding of climate processes, project future climate scenarios, and provide a basis for assessing the potential impacts of climate change. It involves a suite of global climate models contributed by different research institutions and organizations. These models simulate various components of the Earth’s climate system, including the atmosphere, oceans, land surface, and sea ice. The CMIP6 models are used to generate future climate projections under different greenhouse gas emission scenarios, providing crucial information for climate research, policymaking, and adaptation planning.
Under CMIP6, The Shared Socioeconomic Pathways (SSPs) are scenarios developed to represent different possible future trajectories of society, demographics, and economics. They are used in conjunction with Representative Concentration Pathways (RCPs) to explore a range of climate futures. Two specific SSPs mentioned and used in this study are SSP2-4.5 and SSP5-8.5:
SSP2-4.5 (Medium Challenges and Mitigation): This scenario represents a future where global society faces moderate challenges in terms of sustainability and environmental issues. It assumes that, through a combination of technological advances, policy efforts, and societal changes, there is a successful mitigation of greenhouse gas emissions. The radiative forcing associated with this scenario is approximately 4.5 Watts per square meter by the year 2100.
SSP5-8.5 (High Challenges and Limited Mitigation): This scenario portrays a future where global society faces high challenges and experiences limited mitigation efforts. It envisions a trajectory where economic and population growth continues without substantial efforts to curb greenhouse gas emissions. As a result, radiative forcing is projected to reach approximately 8.5 Watts per square meter by the end of the century, indicating a high level of warming and associated climate impacts.
As SSP2-4.5 is considered as the intermediate and most plausible, it was chosen as the first scenario for this study. However, according to recent studies, the world is almost at the limit, if not already exceeded SSP2-4.5 and is heading for the pessimistic SSP5-8.5 trajectory. Accordingly, to be sure that we cover the whole range of possible climatic scenarios, SSP5-8.5 was added to the study. The logic behind this inclusion is that by preparing for the worst (i.e., SSP5-8.5), solutions for intermediate scenario will surely be included. Therefore, the SSP2-4.5-SSP5-8.5 ranges were chosen for the study.
2. Methodology
In this section, the sequential methodology for building the climate indicators is explained:
Step 1: Collection and review of relevant data and material
While open-access databases and global climatic models can be considered as potent inputs for agroclimatic risk assessments, their resolution, even after downscaling, presents certain challenges. At each downscaling step, several assumptions are made. These assumptions are often associated with uncertainties that will ultimately affect the quality of the final output. Therefore, calibration with observed data is often needed to ensure the validity of the approach and the integrity of the simulated results. The CMIP6 (Coupled Model Intercomparison Project Phase 6) model is the most used platform for current climate modeling, characterized by its refined parameterizations and improved representation of Earth system processes. Its integration of complex climate components, including atmosphere, ocean, land surface, and cryosphere, facilitates high-resolution simulations for detailed agroclimatic assessments. In the realm of agricultural studies, CMIP6 outputs serve as valuable resources for quantifying future climate scenarios and their impacts on agricultural productivity. Through its multi-model ensemble approach, CMIP6 enables the evaluation of uncertainty and variability in projected climatic trends, thus enhancing the reliability of agricultural planning efforts. By simulating key agroclimatic variables such as temperature, precipitation, and soil moisture dynamics, CMIP6 helps anticipating potential shifts in crop suitability, growing seasons, and water availability critical for informed decision-making within agricultural sectors. Consequently, the utilization of CMIP6 model outputs stands as a fundamental asset in fostering adaptive strategies aimed at mitigating climate risks and bolstering agricultural resilience in the face of evolving climatic conditions. CMIP6 data was obtained from the COPERNICUS Climate Data Store (CMIP6 climate projections (copernicus.eu)) and from the World Bank Climate Change Knowledge Portal (Data Catalog | Climate Change Knowledge Portal (worldbank.org))
Step 2: Downscaling and correction of climate models
Data is obtained in NetCDF format from both data sources. NetCDFs were then rasterized to be downscaled from 100 km to 1 km. By transposing NetCDFs to raster and then points, the gridded network can be assimilated to a virtual weather station network. Since the different points contain different climatic values, these can be considered as “spatial weather stations”. Through a specific kriging technique using semi variograms, spatiotemporal variations will be revealed. An example is provided below for extreme precipitations in France (Figure 1).
Step 3: Data analysis, exposure methodology, GIS model building
Following the establishment of the climate indicators, exposure is determined. This framework embodies the first elements of the IPCC risk analysis framework (IPCC, 2014; 2020). Exposure is defined based on the occurrence of the different agricultural covers with respect to the climate risk maps. As the studied climatic risks will show spatio-temporal gradients, different crops and livestock types will be affected at different degrees, hence the GIS-based exposure analysis based on their setting with respect to the risk at different horizons. Accordingly, a site-specific approach will be ensured, hence offering a tailored identification of challenges for planning corresponding solutions.
3. Results and discussionsIn Figure 2, the evolution of minimal temperatures across the different horizons and scenarios is presented. The same approach is applied to the list of indicators presented in the introduction section. As can be seen from Figure 2, an increase in minimal temperatures is observed throughout Europe.
The implications of an increase in minimal temperatures in Europe could have profound effects on the agricultural sector. According to the EEA (2019)[1], changes in temperature and precipitation, as well as weather and climate extremes, are already influencing crop yields and livestock productivity in Europe. Weather and climate conditions also affect the availability of water needed for irrigation, livestock watering practices, processing of agricultural products, and transport and storage conditions. For instance, climate change is projected to reduce crop productivity in parts of southern Europe and to improve the conditions for growing crops in northern Europe. Although northern regions may experience longer growing seasons and more suitable crop conditions in future, the number of extreme events negatively affecting agriculture in Europe is projected to increase.
In southern Europe, under a high-end emission scenario, yields of non-irrigated crops like wheat, corn, and sugar beet are projected to decrease by up to 50% by 2050. This could result in a substantial drop in farm income by 2050, with large regional variations. On the other hand, in northern Europe, some of the negative productivity effects caused by climate change could be offset by longer growing seasons and more suitable crop conditions. However, most crops will suffer heavy yield damage in case of drought, frosts, or floods. Grassland is also susceptible to drought, causing cascading impacts on the livestock sector.
In conclusion, the increase in minimal temperatures in Europe could lead to a shift in agricultural practices, with potential benefits in the north being offset by significant challenges in the south. This underscores the need for effective climate change adaptation strategies in the agricultural sector across Europe.
[1] European Environment Agency. (2019). Climate change adaptation in the agriculture sector in Europe. Retrieved from https://www.eea.europa.eu/publications/cc-adaptation-agriculture
In the AgriDataValue project, we anticipated collecting a vast and diverse array of data from various sources and environments. As part of the project’s pilot activities, two SynField smart agriculture systems were installed in different locations within the vineyards of Saint Emilion, France, specifically in areas growing Merlot varieties. Each installation consists of a SynField X5 head node, a SynOdos peripheral node, a meteorological station, soil sensors (measuring soil moisture and soil temperature), leaf wetness sensors, and a pyranometer. The meteorological station records ambient temperature, relative humidity, wind intensity, wind direction, and rainfall. SynField can collect real-time environmental data from the area, providing valuable insights into the vineyards’ conditions.
On June 7, 2024, the National and Kapodistrian University of Athens (NKUA) hosted an Open Day and training session on “Precision Agriculture Technologies” for the local agricultural community. The event highlighted cutting-edge precision agriculture tools and technologies while presenting the impact of AgriDataValue on modern farming practices. Attendees explored advancements designed to improve productivity and sustainability, gaining insights into the transformative potential of data-driven agriculture. The session provided valuable knowledge on how precision agriculture can address key challenges in the industry, fostering innovation within the local agricultural landscape.
In the evolving landscape of agricultural technology, the AgriDataValue project continues to push boundaries, pioneering solutions intended to revolutionize the way we approach farming and agri-environmental monitoring. Building upon the foundation laid out in previous discussions, we delve deeper into the advancements made within the project, focusing particularly on the integration of Federated Deep Machine Learning (FDML) into the platform of platforms being developed within AgriDataValue.
One of the key developments within AgriDataValue is the integration of FDML, an approach that combines the power of deep learning with the principles of federated learning. This integration represents a significant step forward in our mision to address the challenges posed by data scarcity, privacy concerns, and the need for collaborative learning in decentralized environments. In this sense, the new iteration of Hierarchical Federated Learning (HFL) within AgriDataValue offers interesting features, such as the implementation of customizable intra-silo and cross-silo rounds, allowing for finer control over the learning process based on geographic proximity. For instance, fields with closer geographical locations may undergo more intra-silo rounds to refine local nuances, while cross-silo rounds accommodate the integration of diverse data sources from geographically distant locations.
One of the primary motivations behind the adoption of FDML is its ability to preserve data privacy and security while enabling collaborative learning. Farmers can rest assured knowing that their collected data remains within their silos and are not shared with external entities. This not only fosters trust but also encourages greater participation in data sharing, ultimately leading to more robust and comprehensive AI models.
Apart from that, FDML facilitates collaborative learning in decentralized environments, allowing farmers to leverage collective insights without compromising data sovereignty. By pooling together diverse datasets from multiple sources, the FDML framework empowers farmers to obtain more accurate predictions and solutions for their specific use cases. For example, models trained to predict irrigation patterns benefit from a wealth of data originating from various locations, resulting in more robust and reliable outcomes.
The advantages of implementing FDML are manifold. More data in a model translates to more accurate decisions, as the model learns patterns not only from individual farms but also from a broader spectrum of agricultural practices and environments. This is particularly beneficial for applications processing large volumes of geographically dispersed data, where traditional centralized approaches may fall short in capturing the intricacies of diverse farming landscapes.
The integration of FDML represents the commitment of AgriDataValue to innovation and excellence in agricultural technology. By harnessing the power of FDML, AgriDataValue aims to empower farmers, researchers, and stakeholders across the agricultural ecosystem to make informed decisions, drive sustainable practices, and secure a brighter future for food production.