Atmospheric water-vapor signals in GPS data: synergies, correlations, signals and errors

I explore the statistics of integrated water-vapor (IWV) and the ability of GPS to provide information that reflects large-scale weather systems. I demonstrate that the power spectral density (PSD) is a much more robust estimator of IWV statistics than the structure function is for long time scales. I then use the PSD to investigate the IWV statistics of simulated atmospheres. I use a random walk to approximate the Kolmogorov behavior of local atmospheric turbulence. The atmospheres I simulate contain the turbulent behavior sumperimposed on systematic changes to the IWV that might be associated with the passage of fronts. I demonstrate that the PSD associated with such frontal systems might be quite similar to those for pure turbulent behavior. This result may explain why many previous studies that examined PSDs of IWV over a site concluded that the spectra was Kolmogorov. I also derive the first cross-power spectrum of IWV using GPS sites separated by ∼500 km. At this separation, the local turbulence may be assumed to have become decorrelated. I find that even this long wavelength signal consists of components at a wide range of temporal frequencies. The IWV cross power spectrum contains significant energy even at temporal frequencies of 2–4 cycles per hour.