This paper deals with the problem of estimating the number of sources impinging on an array of sensors along with an arbitrary spatially correlated noise field.Under the assumption that the sources are narrow-band and their temporal correlation is longer than that of noise

a new method to solve this problem is proposed which combines the methods of instrumental variable and unitary transformation.In the proposed method

proper instrumental variables are constructed to suppress the effects of noise on the detection performance

and a useful criterion for detection is derived based on the asymptotical analysis of the orthogonality between signal and noise subspace.Detection performance is illustrated by computer simulation and measured data results with comparisons to the traditional and recently proposed methods.