The STP events database is a collection of events relevant to research in the field of solar-terrestrial physics. Although the database contains a variety of datasets the emphasis is on those relevant to the cause or effect of events seen by magnetometers and (imaging) riometers. The database search engine provides a means to show details of events which match certain criteria. Although each dataset can be searched in isolation the true power of the database search engine is correlating between datasets to find events which are cause and effect(s), or related effects. Examples of such correlated searches are:
- Correlate times of substorm onset with riometer absorption events to find absorption events caused by substorms.
- Correlate times of solar X-ray flares with times of with riometer absorption events to find sudden ionospheric disturbances (SIDs) observed by a riometer.
Obviously, the fact that two events happen at similar times does not necessarily mean they are related, it is the responsibility of the scientist to input sensible parameters, and draw sensible conclusions from the results. By correlation we mean two events start and/or end at similar times, not correlation in the mathematical sense (where all corresponding samples are compared).
The database contains many different datasets, e.g., riometer absorption events, times of substorm onsets, solar flares in addition to indices such as Kp and Dst. For the current status please see the database status page. The datasets are divided into 3 categories, events, indices and other criteria. The main properties of the three types are described below. If your browser uses Cascading Style Sheets the descriptions will follow the same colour coding as is used in the search page.
Events are phenomena which happen at uncertain times, with unpredictable magnitudes, such as riometer absorption events, substorm onsets, sudden storm commencements, etc. Some events have a finite duration, and thus have start and end times (e.g., riometer absorption events). Other events, by definition, are instantaneous and do not have either duration or an end time, e.g., substorm onsets or sudden storm commencements.
Unlike events indices have predetermined start and end times, so it is not useful to correlate their times with that of events. They are however useful to provide constraints on the data. Note that an event could correspond to many values in an index. Thus, when using indices as a constraint, some method of reducing the multiple values to one value must be employed. The aggregate function can be minimum, mean or maximum.
The STP events database also contains datasets which do not fall into either of the categories above, and are referred to simply as other criteria, for example SCASI cloud data. When optical instruments are required cloud data is useful to restrict searches to cloud-free periods, but these cloud-free periods are not useful output events.
The search engine requires that at least one events dataset is selected to produce output rows. All the selected datasets can be restricted with some some general search options (start/end dates and time of day, both specified in UTC). In addition, each selected dataset can be restricted with some constraints specific to that dataset. Note that the constraints are cumulative, so in general the more constraints specified the smaller the set of results will be.
The datasets selected to produce output events can also be further constrained with reference to indices and other criteria. Note that including indices and other criteria will increase the search time, so for the fastest search do not include unnecessary options.
The final constraint method (and the most powerful) is to correlate events datasets against the datasets selected as output events.
To successfully correlate between datasets there are a few points to note:
- For those events which do not have endtimes the (start) time is substituted when necessary.
- When correlating between two (or more) datasets please note that only the events within the common periods are candidates for output.
- For some datasets the event times are not entirely precise, but are recorded to a finite resolution (typically 1 minute) so be prepared to examine (and search for) events which may initially appear to show effect just before cause!
- The search speed decreases as the number of correlating datasets is increased (and the number of search results decreases dramatically too). Correlating against start times is faster than correlating against end times.
For more information see the examples page, where various example searches are described, including links to the completed forms.
After a succesful search the output table contains details of matching output events, with one event per row. If indices and/or other criteria were used to constrain the search then the value(s) corresponding to that event will be shown in additional column(s).
The search results are initially sorted by (start) time in ascending
order. The sort field and order may be changed using the icons at the
top of each column. Only the top 200 entries (in the chosen sort
order) are shown at once, though the subsequent pages can be
accessed. It is also possible to download the entire results set as an