Frequency components in barometric pressure data

Some time after getting my weather station and viewing graphs of barometric pressure, I noticed that there was a regular daily pulse in the readings. Curious what other regular periods there were, I decided to take my limited data set (July 2006 through November 2007 recorded once a minute) to see what other frequency components there were. Following are graphs of the most significant results.

I did not use an FFT package to analyze the frequencies since I don't have one at hand. Instead, I rolled my own program that essentially works this way:

Create a set of datapoints where X is the number of seconds since an epoch (midnight, July 1, 2006).
Choose a frequency of interest, and calculate averages for hourly readings modulo the frequency. For a frequency of 24 hours, this would simply be the time of day, for example.
Check if the resulting data contains a pulse. If so, then:
Graph the resulting data using Microsoft Excel.
Repeat with another frequency. For the graph below, I processed frequencies from 3 to 30 hours in 0.1 hour steps, and the rest in one hour steps.

The first graph shows the strength of the pulse by frequency. I also exported my data into a format that could be loaded into a FFT software package. It produced roughly the same graph. I didn't use it, because it's designed for audio data sets, so the units would be confusing. I used it to check that my results were valid.

The remaining graphs show the mean values for frequencies of 24 hours, 91 hours, 177 hours, and 133 hours. A 12 hour frequency was also very strong, but since that's clearly visible in the 24 hour graph, I omit it. The longer the period, the less statistically significant the result, since fewer data points are averaged per graph point. In these graphs, the horizontal axis is the frequency. The vertical axis is sea level pressure.