The performances of high resolution array processing methods are known to degrade in random inhomogeneous media. Such a degradation is caused by random amplitude and phase variations of source wavefronts. In this letter, the popular covariance matching approach to direction finding in the presence of multiplicative noise is extended to the multiple source ease, Using a few unrestrictive physics-based assumptions on the environment, we derive a generic multiplicative noise data model. Based on this model, new direction finding techniques are proposed.