Browsing by Author "Koivunen, Ville"

Technological development has led to a rapid increase of precision agriculture in the last decade. Various sensors can easily be mounted in drones. Monitoring of canopies in different growth stages is therefore quite easily accessible. With this data can different cultivation decisions be made rapidly. In the past remote sensing methods in agriculture have mainly been done with satellite images. The main objective of this research was to determine whether there is a significant difference in accuracy between drone and satellite survey. Hypothesis is that data observed with drone is more accurate, thus there should be noticeable differences in parcel vegetation indices between these two methods. There is a lack of comparative research between these survey methods and usually satellite images have been used only in larger entireties. Two individual parcels used in this study were measured by drones in OPAL-life project in 2016 and 2017. Measuring was made with multi- and hyperspectral cameras and vegetation indices made from these measures were compared with maps made from Sentinel-2 material. Additional comparison was also made between Sentinel-2 based average normalized difference vegetation index and measured grain yield from defined parcels. Results were compared and satellite measurement proved to be quite accurate. NDVI timelines from parcel were almost identical between satellite- and drone images. On the other hand, anomalies and variation in parcel were more observable in drone-based images. Satellite based NDVI pixel values corresponded quite good with drone-based pixel values (R2=0.65). Also, a very significant correlation between vegetation indices and observed grain yields in parcel was observed (R2=0.93) before flag leaf emerging. However the time frame for measurement is very narrow. Results were surprising, but also highlighted the problems involved in this kind of parcel imagery. Satellite images were quite accurate, although some anomalies could not be observed in satellite images. Other issues with surveying formed to be a problem. These were specially the narrow timeframe for measurements, but also clouds were a big obstacle when using satellite images.