Incorporation of computational modules as microservices in the AgDataBox platform and development of the AgDataBox-Map application

Abstract

Digital technologies can provide farmers, specialists, and researchers in the agricultural area the in-deep knowledge of their cropping field, in addition to effective farm management, aiming at increasing profitability and reducing environmental impact. Among the available technologies, there are others, such as precision agriculture, Big Data, Internet of Things (IoT), Unmanned Aerial Vehicles (UAV), robotics, and automation. Digital agriculture requires specific portals and platforms for its adoption. The free web platform AgDataBox (ADB) contributes to farmers' inclusion in the digital agriculture phase. ADB-Map web application and ADB microservices architecture (ADB-MSA) take part on of thematic maps creation (TMs) and the delineation of management zones (MZs) in a friendly and integrated way. Thus, this study aims to integrate the functionalities for creating TMs and MZs through microservices in ADBMSA and incorporating them into ADB-Map application. ADB-MSA provided eight microservices, six of them (statistics, spatial, interpolation, clustering, rectification, and lime/nutrient recommendation) execute procedures based on JavaScript, R, and Python programming languages, while the others are used to store data. ADB-Map was rewritten in JavaScript language and Angular framework, based on a software architecture that decoupled the front-end from the back-end and now uses the resources of ADB-MSA. In the case study, the procedures to create TMs and delineate MZs were carried out satisfactorily with data from a commercial area. Thus, the MZs were generated and evaluated to apply fertilizer according to phosphorus and potassium requirement. In order to improve the interpolator selection process, the new semivariogram model selection criteria were adopted (i) effective spatial dependence index (%ESDI) > 25%, (ii) first semivariance significance index (%𝛾(1)) < 50% and (iii) slope of the model ends index (%SMEI) > 20%, which were applied according to three methods: 1) only with the interpolator selection index (ISI) without application regarding the proposed criteria; 2) the criteria applied after the interpolator selection analysis + ISI, and 3) the criteria applied during the interpolator selection analysis + ISI. Thus, it was observed that, usually, the three methods selected different models and Method 3 was considered the best.