Unsupervised learning techniques provide a way to investigating scientific data based on automated generation of statistical models that describe the data. Because they do not incorporate a priori information, they can be used as an unbiased method to separate data into distinct types. Thus they can be used as an objective method by which to separate data into previously known classes or to find previously unknown or rare classes and sub-classes of data. Hidden Markov models are one type of unsupervised learning method that are particularly applicable to geophysical systems because they include in the model the time relationship between different classes, or states of the system. We have applied hidden Markov models to scientific analysis of seismicity and GPS data from the Southern California region. Preliminary results indicate that the technique can isolate distinct classes of earthquakes from seismicity data, as well as different modes of ground motion from GPS data.