One of the important tasks of crop production management is crop yield forecasting. Currently, remote sensing data such as satellite images and aerial photography are increasingly used as source information for crop forecasting. Due to the rapid development of information and engineering technologies, the use of specialized vegetation indices is also becoming relevant and accessible. The objects of the presented research are experimental agricultural fields located in Leningrad Region. The data obtained on the basis of two polygons with an area of 12 and 28 hectares were used for the work, the growing crop is spring wheat. Test sites were laid on each field - small flat areas with a certain applied dose of nitrogen-containing fertilizers. Aerial photography of experimental polygons was carried out using the DJI Matrice 600 Pro unmanned aerial system (UAS) with a Pika L hyperspectral camera (281 shooting channels in the range of 400-1000 nm). In 2022, wheat samples were additionally taken from test sites simultaneously with the flights and their spectral characteristics were obtained in the fields using a portable laboratory hyperspectrometer. The results of the study demonstrated the advantage of using aerial photography over laboratory hyperspectrometer. Channels from the visible range appear to be the most promising for use in the task of forecasting yields in the conducted experiment, whereas a high multicollinearity of explanatory factors is observed. In addition, a regression analysis was performed. As a result, among the vegetation indices six combinations of spectra were identified for further study in the considered experiment. As a direction for further work, additional field experiments should be conducted and the dataset expanded.