Verification of Earth remote sensing data in order to determine the aboveground phytomass of plantings in the Voronezh Region

The development of Earth remote sensing methods, neural network technologies, creation of machine learning models, etc. will allow developing new algorithms, indicators and criteria that will significantly increase the efficiency of forest monitoring and help reduce financial costs. The basis for the work was the verification of ultra-high- and high-resolution satellite imagery data based on in-situ survey materials conducted on permanent test plots in the Voronezh region. The generated elements of training samples using a neural network classifier allow for automated, highly accu-rate determination of forest stand taxation indicators based on remote sensing data. When verifying the amount of aboveground phytomass calculated based on remote sensing data and in-situ survey materials, significant similarity of the presented results was revealed. In 67% of cases, the average phytomass values determined by different methods did not differ significantly from each other (at a significance level of p<0.05). Reliable differences were found in decid-uous stands with high horizontal canopy density, the presence of a second tier and abundant understory vegetation. As a result of the work, a theoretical basis was laid for further research in conducting remote monitoring in the field of for-est conservation, protection and reproduction. The material of the work is useful in building multidisciplinary practical work in the areas of restoration of biological diversity of phytocenoses, as well as ensuring the integrity and ecological stability of forests, under modern trends in carbon cycles and climate changes.