Optimisation of the correlation between normalised difference vegetation index and sugar beet yield using multispectral remote sensing data

In order to plan harvest times, logistics and campaigns of sugar factories, it is necessary to be able to estimate the yield of the upcoming harvests. In many places, trial clearings are carried out for this purpose. These clearings are done by hand on a few square metres. The coming harvest is estimated on the basis of this. Due to the low sample numbers that can be implemented in this way, only rough forecasts are possible. Modern remote sensing offers the possibility of recording the condition of the beet plants over a wide area. Classifications, which are based on vegetation indices such as the Normalized Difference Vegetation Index (NDVI), allow correlations with yield. This paper describes a methodology based on multispectral orthophotos that makes it possible to achieve very high degrees of correlation using the NDVI in combination with the Excess Green Vegetation Index. The combination of the evaluated unmanned aerial vehicle (UAV) data for both study years over the entire investigation period provides an R 2 of over 0.97. In addition, the transferability to other sensor systems and sensor carrier systems, in particular a manned ultra light (UL) aircraft, is demonstrated. Here, an R 2 of over 0.96 was achieved for both years of the study. In order to reduce the necessary investigation period, various month combinations with the best possible correlation results are also proposed. The period from mid-July to September proved to be optimal here, as this period provided good and stable correlation results. • Using Excess Green Vegetation Index to mask shadows and soil can improve the correlation of NDVI and yield. • The best time for aerial photography is from July to September. • Depending on the ground resolution, the results can be transferred to other sensor systems.