Purpose
This page is part of a study on the temporal clustering of seismic sequences. You — as an expert — are asked to classify a set of event groups into the categories below. The goal is not merely to collect labels: we compare your classifications with a simple automatic criterion in order to measure how closely that criterion reproduces expert judgement, and whether its single parameter is pertinent and well suited to imitate the human choice.
The classes
For each group you are shown a scatter plot of event magnitude versus time. For the most densely populated groups a smoothed event-density curve (left axis) is overlaid as well. Assign the group to one of:
- Single-mainshock — one event clearly dominates the sequence.
- Doublet — two comparable events dominate.
- Multiplet — several events of similar magnitude stand out together.
- Swarm — no event dominates; magnitudes are comparable and no clear mainshock emerges.
- I don't know — use this whenever the plot does not let you decide with confidence. These groups are set aside and do not enter the analysis.
How the criterion works (and what your vote does)
The automatic criterion ranks the events of a group by magnitude and inspects the largest gap between successive magnitudes: its position separates a few dominant events from the background, distinguishing single-mainshock, doublet and multiplet. A single parameter, Δmag, sets how large that gap must be to be considered significant — below it, no event truly dominates and the group is labelled a swarm. Each completed ballot forces the criterion to retune Δmag so as to best fit your own classification. By collecting many ballots and studying the statistics of the fitted Δmag, we aim to assess how well the criterion — and this particular parameter — adhere to expert judgement, and how much the experts themselves agree.
Practical notes
There are 151 groups. Your progress is saved automatically in this browser, so you can stop and resume at any time. Providing a name and affiliation is optional — the study can be anonymous, and these fields are free text used only for bookkeeping. You can submit once you have expressed a judgement (including "I don't know") on every group; the Δmag fit is computed only if you have actually classified at least 121 of the 151 groups (80%).
Questions about the study: Rodolfo Console — rodolfo.console@ingv.it
You can also save an unfinished ballot on the server (button “Save for later” at the top) and resume it later — even from another computer — by entering the same name and affiliation and pressing “Load saved ballot”. A name is required for this; anonymous ballots are still kept automatically in this browser only. The saved copy is discarded once you submit the final ballot.