This paper presents an automated method for creating spatial maps of soil\ncondition with an outdoor mobile robot. Effective soil mapping on farms can\nenhance yields, reduce inputs and help protect the environment. Traditionally,\ndata are collected manually at an arbitrary set of locations, then soil maps\nare constructed offline using Kriging, a form of Gaussian process regression.\nThis process is laborious and costly, limiting the quality and resolution of\nthe resulting information. Instead, we propose to use an outdoor mobile robot\nfor automatic collection of soil condition data, building soil maps online and\nalso adapting the robot's exploration strategy on-the-fly based on the current\nquality of the map. We show how using Kriging variance as a reward function for\nrobotic exploration allows for both more efficient data collection and better\nsoil models. This work presents the theoretical foundations for our proposal\nand an experimental comparison of exploration strategies using soil compaction\ndata from a field generated with a mobile robot.\n