In this paper we analyze five years of recordings (2005–2009) from the National Seismograph Network in the North Island of New Zealand using the power spectral density probability density function (PDF) method of McNamara and Buland (2004). At each station the ambient noise is characterized and the stable noise model is then represented with the modes of the corresponding PDFs over all periods. Obtaining such an accurate long‐term noise baseline for each station provides a reference model that should serve to prioritize maintenance issues for the network operators. The PDF mode low‐noise model (MLNM) for the North Island is then obtained from the minimum of all the noise modes at each period. The maximum and minimum differences between the North Island MLNM and the noise mode model periods are then calculated at each station as a quick assessment tool. The daily and seasonal variations of the noise mode model are then characterized, and the horizontal and vertical mode noise models are then compared at each station. The applied technique is practical for evaluating the cultural noise condition and the earthquake detection capability, as well as the installation design against unwanted tilt and temperature variation.
Online Material: Stable noise models from PDFs, and daily/seasonal model variations.