Spatio-temporal analysis of remote sensing and field measurements for smart farming

For the optimization of crop yield and quality, there is an ongoing development in improving crop management advice, in order to cope with the spatial variability of the growth process, caused by local variations in, amongst others, soil composition, moisture and nutrition content. To achieve this improvement, reliable information is required on the actual status of the vegetation and the expected development and yield given different management scenarios. Remote sensing observations form a valuable information source for assessing the location of suboptimal growth, but hardly ever provide the cause of the arrearage. In order to determine this cause, the observations must be combined with other observations and models and analyzed integrally. This article presents the followed approach and initial results of a pilot project Smart Farming carried out in the Dutch North East Polder. Observations and data from several sources have been collected for a number of potato parcels in 2014. The collected data includes multi-temporal satellite and UAS observations, field based soil, vegetation and yield observations, soil type maps, height maps, historic parcel and crop information and meteorological data. A data driven approach was followed to determine the presence of relations between the various observations in order to couple location and probable cause of sub-optimal crop growth and determine temporal developments in series of observations. The available data was analyzed integrally using correlation, regression and histogram analysis techniques. All resulting spatial layers are visually presented in a GIS based web service environment, so that the advisor or farmer can view the raw and derived information interactively and form his/her conclusions. (C) 2015 Published by Elsevier B.V