Diachuk P. Estimation of carbon sequestered in Scots pine plantations based on forest inventory data and remote sensing technologies

The thesis outlines the challenges which forest inventory faces in Ukraine. Attention is drawn to the need to initiate the systematic implementation of the national forest inventory, to improve forest management and other activities of the State Agency of Forest Resources of Ukraine, to update the database management and mapping software systems utilized by the Ukrainian State Project Forest Management Association. There is an analysis of common data collection and processing approaches based on the field forest inventory and remote sensing data in the thesis, with some focus on the key methods used in this investigation. As well, capabilities of unmanned aerial vehicles (UAVs) to be utilized in forestry are discussed, focusing on the tree height measurement approaches based on the aerial data acquired by quadrocopters. Satellite remote sensing data is an effective tool for estimating the forest ecosystems at the global and regional level, while the UAV data is suitable at local scales. The classification of the satellite image for use in estimating the growing stock volume of stands and sequestered carbon was performed using raster and yaImpute data packages in R environment, while using the Random Forest method to create thematic maps of studied variables. Model covariates were 12 variables including the values of six spectral bands of the Sentinel-2 satellite image with 2A processing level and six respective values of different vegetation indices. A comparison of the estimates of the sequestered carbon in plant biomass within study area, obtained from the analysis of a multilayer raster image using training samples from the data of circular and angle counting sampling sample areas, showed the following results. The stock of sequestered carbon based on the data of circular sample plots was 1.286 Tg C, which was 42% more than the value based on the data of forest management survey (0.905 Tg C), and respective values of carbon stock applying the data of angle counting sample plots was 1.482 Tg C, which was 62% more than respective value from forest management survey data. The deviation between the estimates according to the circular and angle counting sample plots was 15%. A comparative analysis of the estimated of the total growing stock volume within the study area calculated by different methods showed significantly smaller deviations than the value of sequestered carbon in plant biomass. According to local forest management survey data, the total growing stock volume of stands of the forests within study area was 2,685.80 thousand m3. During the analysis of the raster using a training sample based on the data of circular sample plots, the respective growing stock volume value was 2,888.97 thousand m3. This is 7.6% more than the value obtained from forest management survey materials. Analyzing the rasters resulted from the data of angle counting sample plots, the stock was 2,508.47 thousand m3, which is 6.6% less than respective local forest management survey data. Therefore, the general deviation of the estimated total stock was within ±10%. The thesis presents the results of modeling the forest stands parameters of trees based on the nonparametric methods, also the comparison of estimated parameters of forest growing stock volume and sequestered carbon between the statistical forest inventory and forest management survey data. In addition, the possibilities of photogrammetric processing of images obtained by UAVs to identify trees and measure their heights in a mature Scots pine stand are examined.