In order to sort frequency-hopping networks, identify and track signals effectively, an novel joint estimation algorithm based on spatial polarimetric time-frequency analysis is proposed as the solution for the two dimensional direction of arrival (2D-DOA) and polarization estimation of Frequency Hopping (FH) signals in underdetermined condition. First, the data model for polarization sensitive array of FH signals is built. On this basis, the SPSTFT, the extendable form of spatial polarimetric time frequency distributions, is derived. Meanwhile, a new combined time-frequency(TF) analysis method is presented to locate the auto term region of each hop in the TF plane. Consequently, the SPTFD and SPSTFT matrix for each hop is generated. Finally, depending on the spatial and polarization information getting from SPTFD and SPSTFT matrix, the 2D-DOA and polarization are estimated by two different methods. With splendid estimation precision, this novel method requires less computation burden without multiple-parameter optimization and pair-matching. Simulation results show that the proposed algorithm could estimate 2D-DOA and polarization in underdetermined condition effectively, and the SPTFD matrix method has high estimation accuracy.Besides, the collision hop could be processed.