Application of big data analytics and artificial intelligence in agronomic research

Abstract

Agronomic research involves study of crop-soil-environment interactions validated by field experiments. Modern statistical tools complemented in designing field experiments and immensely contributed in drawing useful inferences for developing good agronomic practices for increased crop production, input-use efficiencies and environmental sustainability. However, timely analyses of huge agronomic data sets having huge spatio-temporal variations are important in translation of research to real field situations. In order to enhance the reach of agronomic research, use of emerging tools of big data analytics, geo-referenced satellite information Unmanned aerial vehicle (UAV) based imaging and artificial intelligence (AI)-based techniques which can process large data sets are described which could be validated by agronomic field experiments. Some areas of data science for use in agronomy include: satellite and UAV based data acquisition, Internet of Things (IoT), AI (machine and deep learning) and big data analytics. Recent studies demonstrated that using AI-based algorithms the accuracy in yield prediction and image classification is enhanced up to 85%. Our study using all India wheat (Triticum aestivm L.) production data for a period of 58 years showed that Bi-directional Long Short-Term Memory (LSTM) model reduced error in time-series prediction to the order of 50% in comparison with conventional statistical models. Since agronomists aim for holistic understanding of agro-ecosystems, System Dynamic Model (SDM) of specific agricultural systems wherein the topography, climate, hydrology, natural resources and societal requirements are coupled along with their feedback impacts on agronomic systems has been introduced and discussed. Agronomists could collaborate with experts in other fields such as data science to bring about a new paradigm agronomic research