In many signal processing applications high accuracy signal parameter estimation from sensor array data is a significant problem. Much of the recent work in array processing has focussed on methods for high-resolution location estimation. Model based estimation techniques require accurate knowledge of the so-called array manifold. In practice, the array response is often determined by measuring the array response when only one emitter is radiating and the signal parameters of which are allowed to vary in a known way. This paper addresses some of the practical issues that arise in generating the socalled array manifold from a finite collection of calibration vectors. For high-resolution signal parameter estimation techniques to be successful, the interpolated array manifold has to satisfy certain smoothness conditions. A paradigm for generating an array model from noise corrupted calibration vectors is developed. The key idea is to use a local parametric model of the sensor responses. The potential improvement using the suggested scheme rather than an ideal array model is demonstrated on real data collected from a full-scale hydro-acoustic array.