Day 4

Detailed paper information

Back to list

Paper title Measuring Apparent Agricultural Field Heterogeneity – a comparison between UAV and satellite
Authors
  1. Quirina Merz ETH Zurich Speaker
  2. Achim Walter ETH Zürich, Institute for Agricultural Sciences, Group of Crop Science
  3. Helge Aasen Agroscope
Form of presentation Poster
Topics
  • C4. HAPs/UAVs
    • C4.01 Innovative UAV applications
Abstract text Agricultural fields are seldom completely homogenous. Soil, slope, and previous management decisions can influence the conditions under which a crop grows and determine its nutritional needs. However, in current farming situations in Switzerland, fertilizer is still spread mostly subjectively, according to the knowledge of the field manager. It is crucial that fertilizer is applied at the right time and in the right place. This prevents over-fertilization of the field, fertilizer run-off, and saves fertilizer. Variable rate technology (VRT) can help to apply fertilizer according to the actual need of the plants. VRT can be based on field imagery as input for fertilizer calculation - this imagery can be obtained with hand or tractor mounted sensors, UAVs or satellites. However, VRT in combination with sensors is very expensive. It is estimated that the use of VRT and sensors only pays off once a certain threshold of heterogeneity in the field is reached. The profitability of VRT systems also varies depending on the cost of sensor technology which is used. UAV-based field imagery is available at a very high spatial resolution of a few centimetres, but the costs for the flight missions are considerably high. Satellite-based data comes at little or no cost at all, however, the spatial resolution is much lower, this can cause errors especially in small scale fields. Overall, data on field heterogeneity is scarce, especially in the context of spatio-temporal changes throughout the vegetation season. Further, it is unclear, which spatial resolution is needed to capture the in-field variability reliably in small scale fields. In this contribution, first results of comparing spatio-temporal dynamics of field heterogeneity between high spatial resolution and low spatial resolution are introduced. A fixed-wing UAV (WingtraOne) was regularly flown over a small rural area in Switzerland at relevant times of the vegetation period over 2.5 consecutive years. Fixed-wing UAVs manage to cover 50 to 100 ha in one flight and are thus ideal for these studies. The study area included a diverse set of crops, ranging from winter wheat, canola, maize, sugar beet, sunflower to grassland and vegetables. The drone was equipped with different cameras: a high-resolution RGB camera (Sony RX1RII, 42 megapixels) and 3 different multi-spectral cameras (Red-Edge M, Red-Edge MX and Altum, all by Mica Sense). All multispectral cameras captured data in at least 5 bands of the RGB and near infrared spectrum (Altum also collected thermal data), which were used to calculate vegetation indices to assess crop health status. The spatial resolution of 0.7 to 1.2 cm (RGB) and 6 to 8 cm (multispectral) offered a very highly resolved dataset which was then used to investigate the field heterogeneity on various spatial scales. Soil maps and field book data of the respective farm managers complemented the dataset.