Geomagnetically induced currents (GICs) can be driven in terrestrial electrical power grids as a result of the induced electric fields arising from geomagnetic disturbances (GMD) resulting from the dynamics of the coupled magnetosphere-ionosphere-ground system. However, a key issue is to assess an optimum spacing for the magnetometer stations in order to provide appropriate monitoring of the GIC-related GMD. Here we assess the vector correlation lengths of GMD and related amplitude occurrence distribution of the variations of horizontal magnetic field $dB_{H}/dt$. Specifically, we study the GMD response to two storm-time substorms using data from two magnetometer arrays, the Baltic Electromagnetic Array Research (BEAR) Project in Scandinavia and the Canadian Array for Realtime Investigations of Magnetic Activity (CARISMA) array in North America, so as to determine the optimal magnetometer spacing in latitude and longitude, for monitoring and assessing GIC risk. We find that although magnetic disturbances are well-correlated up to distances of several hundred kilometers at mid-latitudes, the vector correlation length rapidly drops off for station separations of less than 100 km within the auroral oval. In general geomagnetic fluctuations are stronger and more localized in the auroral zone. Since the auroral oval is pushed equatorward during intense magnetic storms, we highlight that networks using a station separation of $\sim 200$ km should provide an excellent basis for monitoring both small and large scale geomagnetic disturbances. A monitoring network with this station spacing is recommended as being optimal for assessing the role of GMD in driving GICs in the electric power grid.