Abstract: Root morphologies can dominate the plant growth to absorb the water and nutrients. It is also required for the patterns of root morphology in the plant adaptation to ecological environments. However, some challenges are posed from the invisibility of plant root systems buried underground during the quantitative analysis of root morphologies. This study aims to achieve the non-destructive interpretation of root morphologies and root biomass. A quantitative analysis was also proposed on the root morphologies and root biomass using ground penetrating radar (GPR). A simple approach was provided to non-destructively reveal the growth form and spatial distribution of underground tree roots. The nine-year-old Chinese ash tree (Fraxinus chinensis Roxb.) was taken as the research subject in northern China. Initially, a 900 MHz GPR system was employed to conduct a square grid scan of the root system. A GPR C-scan data volume was generated with the root morphologies. Subsequently, the vertices of the root system reflection hyperbola in each B-scan image were extracted to spatially locate the root systems within the C-scan data volume. Additionally, the total propagation time of radar waves through the root systems was derived from the A-scan images intersecting the root system hyperbolic vertices. The root system diameter was then estimated after operation. After that, the localization of root system was facilitated to reconstruct the three-dimensional (3D) spatial distribution of the root system architecture (RSA). Finally, the root systems were modeled as the frustum-of-a-cone, in order to quantitatively analyze morphological characteristics, such as the length, surface area, volume, and spatial growth angles. A relationship model was established between the root volume and biomass. Non-destructive estimation of root biomass was realized using root volume characteristics. The feasibility and accuracy of the quantitative analysis was also validated for the root morphologies and root biomass. The whole root excavation was evaluated after validation. The research results indicated that: (1) The GPR was effectively detected the buried root systems, with a correct identification rate of 64.2%, primarily located at the depths of 0.1 to 0.4 m (76.9%); (2) The accurate locating root system was achieved in the vertices of the reflection hyperbola from the GPR C-scan data volume. There was an average root point localization error of 15.8%; (3) The diameter of root system was effectively estimated, according to the total propagation time of radar waves, indicating an average estimation error of 26.4%; (4) The quantitative analysis was feasible on the root morphologies using the frustum-of-a-cone model. The total length, total surface area, and total volume of the RSA reconstructed by GPR were 786.5 cm, 12588.9 cm², and 19969.9 cm³, respectively, with the estimation accuracies of 66.77%, 80.72%, and 93.84%, respectively; (5) The root volume was effectively represented the root biomass, with an estimated biomass of 0.131 g/cm² and an accuracy rate of 76.09%. The root-soil mechanics can also be expected to non-destructively monitor the root growth. Additionally, the dynamic assessment of root biomass accumulation was enhanced to understand the root growth and survival mechanisms. The great contribution are gained for the ecological adaptation strategies of the roots in complex environments. The finding can also provide the technical support for the forestry production and ecological protection.