Abstract <p>In the ecotone of the upper tree line, the relationships between the values of biometric parameters of Siberian larch (average radius of the horizontal projection of the crown, the diameter of the tree at the root collar and its height) were studied using ground-based measurements on circular sample plots and ultra-high spatial resolution aerial photographs obtained by an unmanned aerial vehicle. A nonlinear regression model, a model using the random forest method, and an ensemble of models using machine learning methods were created to establish a relationship between the values of the root collar diameter and the crown radius of a Siberian larch specimen. The obtained models have a high level of adequacy both qualitatively and quantitatively (<i>R</i>² more than 0.95) levels. The predictive capabilities of the nonlinear regression model outside the training set were better than those of the machine learning models, so it was used together with allometric equations to quantify the phytomass of Siberian larch based on the root collar diameter and carbon sequestration in the study area using data obtained from crown decoding of 88.6 thousand Siberian larch specimens. It was found that in the upper treeline ecotone over an area of 7.32 km² the value of aboveground and underground phytomass of Siberian larch is 1355.2 tons of dry mass, which contains 677.6 tons of carbon, or 2484.5&#xa0;tons of CO²-equivalent.</p>