Modelling forest biomass dynamics in relation to climate change in Romania using complex data and machine learning algorithms

Forest biomass controls climate stability, many ecological processes and various ecosystem services. This study analyzes for the first time the recent changes (1987-2018) of forest above-ground live biomass (AGB) in Romania, based on a complex volume of remote sensing and forest inventory data that were modelled yearly using a series of sophisticated statistical algorithms. Subsequently, after modelling interannual AGB data, yearly raster values (similar to 2 billion total pixel values) were explored as trends over the 32 years, using the Sen's slope estimator and Mann-Kendall test. A large volume of climate data was also processed in this study, in order to detect possible statistical relationships between climate and forest biomass, after 1987. Results showed a mean multiannual value of forest biomass of similar to 185 t/ha and a total AGB amount (stock) of about 1.25 billion tons (similar to 1249 million tons or megatonnes/Mt) across Romania. Regarding forest biomass changes, findings revealed increasing and decreasing AGB trends that account for similar to 70% and 30%, respectively, of the countrywide forest biomass changes. However, it was found that about half (similar to 48%) of all positive AGB trends are statistically significant, while negative AGB trends have a statistical confidence on only one-fifth (similar to 21%) of their spatial footprint in Romania. Overall, upon averaging and summing up all statistically significant values of positive and negative trends, an average AGB increase of similar to 3 t/ha/yr and a total forest biomass gain of similar to 205 Mt were found in Romania, over the entire 1987-2018 period. The various regional statistics highlight a more complex picture of AGB changes across the country. The analysis of interannual eco-climate data indicated a low to moderate climate signal in AGB changes, revealing that climate change is not a major driving force of AGB dynamics, at least according to the data and methodology applied in this study. The results can be useful to governmental forestry, climate and sustainable development policies in Romania.