Data Fusion in Agricultural Information Systems

Abstract

Data has an increasing role in agricultural production and management at all scales depending on raising importance of yield efficiency and sustainability. Remote sensing systems provide spatial information at some observation instants and real-time terrestrial monitoring systems provide temporal information at some observation points. Data fusion methods appear as feasible way of multi-temporal mapping of information in Agricultural management. Data fusion uses computational models and machine learning methods on available spatial, temporal, and multi-temporal data sets. In this chapter, basics of data indexing and segmentation in Agricultural monitoring is given in accordance with application examples of time-delay neural networks, convolution and the wavelet transformation for data fusion. Frequently used agro-meteorological indices and yield efficiency relationships are also explained. Since many of the required monitoring parameters are usually not feasible for real-time data acquisition, data fusion methods enable estimated parameters indirectly from the correlated set of available data. In contrary to distributed characteristics of data resources and the users in agriculture, computational systems have centralization trend through “Data as a Service” (DaaS), “Platform as a Service” (PaaS) and “Artificial Intelligence as Service” (AIaaS). Data fusion is especially expected to have an increasing role for large scale, continuous-time data services in Agricultural applications.