An efficient sampling strategy should address knowledge gaps, rather than exhaustively collect redundant data. In this study, spatial uncertainty in DEM estimates was used to locate targeted sampling areas in the field. An agricultural vehicle equipped with RTK-DGPS was driven across a 2.3 ha field area to measure the field elevation. Data were collected at 3.05 m (10 ft) intervals in a continuous fashion at a speed of 9.6 mph. A geostatistical simulation technique was used to simulate field DEMs with different measurement pass intervals and to quantitatively assess the spatial uncertainty of the DEM estimates. The high uncertainty areas for each DEMs were classified using image segmentation methods and targeted sampling was performed on those areas. The resulting DEMs were compared with each other to evaluate the effect of including targeted measurement on DEM accuracy. The addition of targeted measurements significantly reduced the time dedicated for the re-sampling effort and resulted in DEMs with lower RMSE. For the widest interval between sampling passes, the RMSE of 0.46 m of the DEM was reduced to 0.25 m after adding the targeted measurements which was close to the 0.22 m RMSE of DEM with whole field re-sampling. The results show that spatial uncertainty models are useful to design targeted sampling for field mapping. The method is not limited to map elevation data but can be extended for mapping other spatial data.