The semivariogram can provide a description of the spatial correlation of a variable that represents a continuous function of space co-ordinates. Therefore, it is not only useful to improve sampling strategies and validate the collected data, but it can also define the main parameters that are involved in most mapping algorithms. The spatial correlation may, however, change depending on the values of the variable, which is not without consequences on the quality of the maps of radioactive deposition. This work investigates the spatial correlation for different levels of 137Cs deposited in Austria. It is shown that low levels (< 5 kBq/m2) and high levels (> 40 kBq/m2) of radioactivity present less correlation in space than the mid-levels (5-40 kBq/m2). As a result, maps establishing the probabilities to be under or above a specific deposition threshold will have an uncertainty associated to the estimated values that is directly correlated to the different levels of radioactivity.