The ground surface of the Earth is usually mapped using microwave radar, and in particular synthetic aperture radar. A different technique is presented for the imaging of isolated ground surfaces, which may range from the order of a metre to a few kilometres in diameter, using radiowaves generated from antennas on the ground surface of the Earth. The technique is first formulated using groundwave propagation theory to solve the forward electromagnetic scattering problem of the isolated surface. Conjugate gradient methods are then used to solve the inverse problem to obtain the normalised surface impedance distribution in an iterative manner, with the establishment of a cost function of the measured and scattered fields around the surface. Numerical simulations at 0.5 MHz of two ground features, namely dry ground and dry–wet mixed ground surfaces confined within a diameter of 500 m and surrounded by seawater, show that the distributions of both real and imaginary parts of the normalised surface impedance can be well reconstructed using the technique. It is also shown that the Polak–Ribiere–Polyak conjugate gradient method gives a faster rate of convergence than the Fletcher–Reeves method. The simulation at 300 MHz of dry ground with a buried metallic object confined within a diameter of 50 cm shows that the metallic object and its position can be identified from the images of normalised surface impedance distribution. The study demonstrates a new technique for the tomographic imaging of isolated ground surfaces which may be useful in ground surface mapping, remote sensing and monitoring, and detection of buried metallic objects.