Distributed-acoustic-sensing (DAS) data have been widely used to monitor multifracture hydraulic fracturing. Interpreting hydraulic fracture geometry (width and length) using DAS data is a popular topic of current research. However, the previous study can only estimate the fracture width near the offset well from single low-frequency DAS (LF-DAS) data. Due to the multiplicity problem, no study was attempted to characterize fracture lengths from DAS data. In this paper, we propose a new model to inverse the fracture length and width over treatment time from multiple sources of data [LF-DAS, high-frequency DAS (HF-DAS), and injection rate data]. First, HF-DAS waterfall plots of the treatment well and injection rate curves are aligned to determine the volume of fluid injected into each fracture. LF-DAS data along the offset well are related to fracture width by the Green function. The fracture length and width are determined by combining the Green function and fluid volume constraint. Second, the Picard and the least-squares methods are used to improve the robustness of the model computation. The inversion model is validated by a fracture propagation case generated by the displacement discontinuity method (DDM). In addition, the effects of the distance between the fiber and the wellbore, spatial sampling spacing, and fracture spacing on the computational stability of the inversion model are discussed. By combining DAS data and other monitoring data (e.g., inject rate) from treatment or offset well, the fracture length and width with treatment time can be accurately estimated. Results of fracture geometry interpretation can optimize fracture design and help improve production efficiency.