A new subspace tracking algorithm based on instrumental variable is proposed to estimate the two dimension（2D） localization of aiming at tracking problem of spacial moving nodes in colored noise.By using the rank-one update model

two unconstrained cost functions are established

and signal subspace is derived through the recursive least square solution of the cost functions.Using this derived signal subspace

construct matrix equation and through lease square solution realize direction tracking of spacial moving node

the obtained 2D spacial angle can be paired automatically.In the case of arbitrary planar sensor array

the proposed algorithm and the well known EIV-PAST have good node tracking characteristics

signal subspace angle

signal subspace error are small

the orthonormality performance of proposed algorithm is much better than EIV-PAST algorithms.