Developing and evaluating an aerial sensor platform (ASP) to collect multispectral data for deriving management decisions in precision farming

Abstract

Precision farming aims at managing and adapting the application of input factors to the current demand of the plant, based on the given growing conditions within a field. In order to enable suitable site-specific management strategies, it is necessary to identify the cause for a given variability within the field. While tractor based sensor systems are already implemented in agriculture, use of aerial sensor platforms (ASPs) such as unmanned aerial vehicle or unmanned aircraft systems (UAV/UASs) are in development and hence still require application based testing. In this study an ASP was tested and evaluated concerning its suitability to follow a given route within a field, and to collect georeferenced multispectral data which can be used for deriving management decisions (as fertilizer application) in precision farming. The results showed that the ASP reached and maintained the target altitude during the flights, along with achieving deviations within a limited range of the relevant position parameters (pitch, roll), and this appears to be acceptable for agricultural purposes, as for example fertilizer applications. The analysis of collected multispectral data did indicate only slight correlations between plant characteristics (biomass, leaf area index, nitrogen content, corn yield) and indices derived from multispectral data. However, the ASP itself could serve as an aerial sensor platform for deriving management decisions in precision farming, as long as an approximate number of about 35 measurements per ha represents a sufficient database for deriving management decisions. In this study the agricultural machinery used had a working width of 6–12m without additional possibility of section control. Thus, the possibility to adapt the application of input factors (as for example fertilizers) spatially within the field was limited to the given working width of the spreader. In this case, the given spatial resolution of aerial data (with about 3 measurements per grid) can still be used to derive site-specific fertilizer management prescriptions. Due to technical limitations of the used autopilot, exact flights covering determined routes in cm or m level accuracy were not possible in this experiment, but might be possible with other autopilot systems. Existing problems with the quality of multispectral data can be solved by installing a lightweight construction reference system in the ASP or integrating a sufficient reference system on the ground. Considering these improvements the ASP could be used as suitable sensor platforms to collect multispectral data, which can be used for deriving management decisions in terms of precision farming.