Elimination of the baseline: How to find background "noise"

A seismograph generates pixels on the page simply by drawing a line across the page. Even if there is 0 tremor, there still would be a line and the ImageJ software will count pixels. Further, the page itself contains division lines for each minute and you may also be measuring the division lines between rows. Subtraction of these pixels is important since they do not represent tremors.

It is not overly important that you get this part right if you are just comparing two seismographs since both of your raw numbers should be offset by the same amount. You can run using the numbers found here, or make your own numbers in a similar way.

You have options. You can just subtract the pixels from a blank page, or you can attempt to find a quiet seismograph printout with almost no tremor.

Subtracting a portion or entire blank page:
Here is what a blank page looks like for a 6 hour graphic.

I have given you the entire page above in case you want to look at seismographs one entire page at at time. There is something to be said about being lazy. The good thing about looking at whole pages is all the pixels are counted and you get an overall value quickly. The downside, is you don't know which 15 minute period was highest. At the bottom of this page I have included 12, 24, and 48 hour blank pages.

For a 6 hour page, baseline values to use are:

• Mean 4.859 for the full page with an area of 1090800 pixels.
• Mean 3.733 for a 15 minute segment with an area of 34500 pixels*

*Note: The 34500 pixel area includes both top and bottom dotted lines but only one of the two side continuous line. When using the either constant, make sure the area you analyze equals the area for the constants you use.

Using a quiet period for a baseline
Find what appears to be a perfectly quiet time for the type of printout you are working with and then run an analysis for the same area you use to take samples.

On an island with an active volcano that might be easier said than done. After some searching I located a seismograph on the NW slope of Manua Kea.

Notice right near the bottom of this page there is a quiet period. An analysis of a 15 minute window of this quiet time yielded:

Mean 32.27 for a 15 minute segment with an area of 34500 pixels.

Next up, subtracting baseline error to obtain net tremor activity.