We analyse the estimation of values of a survey variable throughout a continuum of points in a study area when a sample of points is selected by a probabilistic sampling scheme. At each point, the value is estimated using an inverse distance weighting interpolator, and maps of the survey variable can be obtained. We investigate the design-based asymptotic properties of the interpolator when the study area remains fixed and the number of sampled points approaches infinity, and we derive conditions ensuring design-based asymptotic unbiasedness and consistency. The conditions essentially require the existence of a pointwise or uniformly continuous function describing the behaviour of the survey variable and the use of spatially balanced designs to select points. Finally, we propose a computationally simple mean squared error estimator.