Abstract : The purpose of this study was to determine the amount of predictable noise at Uinta Basin Seismological Observatory (UBO) by the development and application of noise prediction filters. A comparison of the surface and the 200-ft buried array was conducted to determine if a large amount of noise was present which attenuates rapidly with depth. The determination of the amount of predictable noise at a particular array is desirable to determine or to explain the effectiveness of multichannel processing. A measure of the amount of predictable noise at a particular station gives a rough measure of the signal- to-noise improvement possible with multichannel processing above that obtained with straight sum processing. In the presence of random noise, both processes give a maximum signal-to-noise improvement of the square root of N, where N is the number of seismometers in the array. Twenty-two noise prediction filters were designed on, applied to and evaluated for the surface and subsurface planar arrays at UBO. Prediction filters were designed upon the nine outlying individual seismometers in each array and also upon the nine sensors summed into three rings. Each type predicted the ouput of the center seismometer.