The success of low frequency telescopes like EVLA, GMRT, LOFAR, LWA and MWA heavily depends on the ability to calibrate the instrument to high accuracy. The ionosphere causes spatially variant phase distortions on passing low frequency radio waves, degrading the performance of conventional methods like self-calibration. The single existing implementation of a spatially variant calibration scheme, the field-based calibration technique by Cotton et al. (2004), shows a significant improvement over self-calibration, but is by design limited to array sizes smaller than the scale size of dominant ionospheric fluctuations. We present a new calibration method which intrinsically does not have this limitation. Our implementation, named S.P.A.M. (Source Peeling and Atmospheric Modeling), uses phase calibration solutions towards several bright sources within the field-of-view to constrain an ionospheric phase screen model at fixed height above Earth's surface. While imaging, the model predicts phase solutions in arbitrary viewing directions. Application of SPAM on selected test fields from the 74 MHz VLA Low-frequency Sky Survey (VLSS; Cohen et al. 2007) shows a 30-50% reduction in the background noise and better focussing of point sources as compared to self-calibration and field-based calibration. SPAM is currently expanded with a multi-layer phase screen model, and tested on 150 MHz GMRT data and 74 MHz VLA-A data, with positive initial results. Although SPAM is in many ways a pilot project, it may prove to be useful for the future development of more advanced calibration algorithms.