Class EventMetadataLookup

Lookup the event metadata associated with an OpenQuake quake (primarily for probabilistic modelling).

There are two datasets in the HDF5 file of interest here: - /events : we first need to lookup the event metadata, specifically the ID of the rupture associated with it. - /ruptures : this holds the details of the quake rupture that corresponds to the event. Specifically, we're interested in the number of occurrences ('nocc') and the rate of occurrence ('occurrencerate'). We can then use these details to eventually calculate the probability of the quake .

Each quake event has both a seismic source and a rupture. The source is the geographic location of the quake, whereas the rupture represents a quake of a given magnitude for a seismic source. The same source can correspond to many different ruptures (i.e. different magnitudes), and the same rupture can correspond to many different events (as we're modelling the uncertainty of a quake of any given magnitude). The number of occurrences of a given rupture is stored as 'nocc'. E.g. a rupture with nocc=5 will have 5 different events associated with it.

There are two different 'flavours' of sources. These are defined by the 'uncertaintyModel' in the OpenQuake job's input files (the 'uncertaintyModel' is defined in the 'sourcemodellogictreefile' XML file). - characteristicFaultSource: these have a fixed magnitude, and a fixed 'occurRates' (which maps directly to the 'occurrencerate' in the resulting HDF5 file). - pointSource: these have a range of magnitudes. With a large number of SES, each magnitude in the range ends up getting sampled by an event. In these cases, the 'occurrencerate' gets calculated by OpenQuake.

Note that these probability calculations are designed for a 'patched' version of OpenQuake where the 'n_occ' values can be overridden manually when generating the HDF5 file. This changes the model to be based on weighted events, rather than a straight SES event-based model.