The Rao efficient scores test is an asymptotic optimal detector to test weak signals in colored non- Gaussian background.Since applying technologies of prewhitenning and gaussianization

it can get better detecting performance than conventional match filter.Gaussian mixture autoregressive model is addressed with the test problem being described.Estimations of parameters for this model are required for construct of the detector.Based on these parametric estimations

the filters of prewhitenning and gaussianization are applied in constructing the practical modular Rao efficient scores test.With its detecting performance being analyzed deeply

the causes why technologies of prewhitenning and gaussianization can improve detecting performance are brought to light.Finally

a test instance with lake trial data is illustrated.