Objectives of the service
The general level of digitization in the agrifood sector is growing, but it is still at an early stage. Many farmers rely extensively on past experience and traditional technical knowledge for farm management, especially for essential decisions such as irrigation and field yield estimation. However, this management system has well acknowledged limitations, which become even more relevant in presence of large heterogeneous fields and a changing climate. Stakeholders, among which farmers, recognises that a more digitized agricultural sector is urgently needed to overcome these limitations.
The YieldOptimizer service, designed as an extension of the Agricolus cloud-based platform, addresses these needs by providing yield and water stress variability maps. The two separate maps are based on algorithms able to predict in advance high/medium and low potential yield areas and areas subject to high/medium/low water stress. Farmers and other actors of the sector will be able to better plan irrigation actions and farm management in general, thanks to data based and reliable previsions. This will foster an important improvement in field management, crop cycle planning, use of scarce resources such as water, and overall sustainability.
Users and their needs
User engagement activities are deemed essential to develop a successful solution. Therefore, considerable time was devoted to interviews with different groups of users such as farmers working on farms of different extents, consultancy companies specialized in the agricultural sector, and companies supplying agricultural products.
Faced with an ever-changing environmental and economic scenario, farmers find themselves increasingly in need to integrate their traditional knowledge with the use of innovative technological tools. In this regard, YieldOptimizer addresses their needs for constant and efficient monitoring of the crops’ health status, the need to plan crop operations effectively, to intervene timely in presence of water stress, and their need to plan economic investments based on a reliable expectation of yield levels.
The main challenge is to find the best possible balance between effectiveness of the information given and usability. For this reason, specific attention is given to the robustness of the tool, the automation of the process, and the usability of the information provided to the end user.
Service/ system concept
YieldOptimizer is a service integrated with Agricolus Farm Management System architecture. The solution offers two models for yield and water stress prediction zones for generic arable crops, yield zone, and water stress zone.
The yield zone forecast model predicts variability of the yield on the field in the form of classes (high/medium/low yield areas), using a machine-learning approach and trained with multi-temporal satellite and yield datasets. The yield zone forecast model allows the farmers to customize the crop operation in different zones of the field during the season, and timely plan the tasks.
The water-stress zone forecast model is a satellite-centred solution which relies on water stress indices on bare soil calculated from the satellite data. The model provides a water retention map of the field showing the ability of each area to keep a high level of humidity over time. The map is then classified into three classes (low, medium, high potential water retention).
Space Added Value
The two complementary models are based on the use of Earth Observation satellite data, the heart of the Yield Optimizer service. The resolution of satellite data allows accurate detection of in-field differences and identification of homogeneous areas as well. The satellite data are provided with high frequency enabling the forecasts to be continually updated.
The satellite data available all over the world make it possible to extend the service to different world areas, and to take advantage of multi-year historical data.
Current Status
The Yield Optimizer Kick Start activity after 6 months from its Kick Off is being successfully concluded.
User needs (expressed by customers located in Europe) were successfully collected (via virtual meeting) and consolidated to define the service requirements.
The technical feasibility of all the fundamental building blocks, which constitute the architecture of the system, was examined and the prototype models were tested successfully.
The commercial viability of the proposed service was analysed and confirmed.
