1.- Review of pending requests: =============================== 1.1.- Request #128 (S. Zuzlewski): ---------------------------------- --> Allow NULL for Amp.datetime. ------------------------------------------------------------------------ ==> Value could be set to: - NULL: problematic because Caltech is using this field for partitioning the Amp table. - Some bogus value. - Origin Time. * Bring issue to Standards Group: Currently in the NC & SC databases, amplitude times are only approximation; you cannot rely on the Amp.datetime field. They might contain miscellaneous values such as P pick time, S pick time, event origin time, or some time within the event waveform packet (which describes the situation with currently imported spectral acceleration values from CISN partners). What this means is that the real amplitude time could be off by as much as 20 minutes in certain cases. The WG considered just changing the constraint on the Amp.datetime field and allow for a NULL value whenever the amplitude time is unknown. However, this causes database administration difficulties for Caltech. QUESTION: Is the fact that the amplitude times in the database are not reliable acceptable to the seismologists? After hearing some feedback from a few seismologists, the group was under the impression that not many are actually interested in this parameter. If the answer is NO, i.e. we want to know whether the time is "correct" or "unknown", we need to either DELETE the "unknown" times or add additional information to the time observation that qualifies it. Would it be acceptable to REMOVE times that are thought to be "unknown"? Current amplitude times that are "unknown" are from historic data: - NC WA amplitudes readings from pre-CISN software. - ALL CISN SGM spectral acceleration times imported from CISN partners. ------------------------------------------------------------------------ ------------------------------------------------------------------------ Amplitude times in the database are only approximations and therefore are not reliable. Affected readings are: - NC WA amplitudes readings from pre-CISN software. - ALL CISN SGM spectral acceleration times imported from CISN partners. The SCWG requested input from the Standards group on the following questions: 1.- Is the fact that the amplitude times in the database are NOT reliable acceptable to the seismologists? 2.- If the answer is no, would it be acceptable to REMOVE times that are thought to be "unknown"? Peggy H. & Dave O. agreed on the following answers: 1.- NO. 2.- YES. Based on this preliminary feedback, the SCWG discussed possible ways to implement this scenario and came up with three options: 1.- Allow NULL values in the Amp.datetime field to represent unknown times. This causes oracle administration issues at Caltech because they are using this field for partitioning the Amp table. Also with this option, the approximate times are lost. 2.- Add a new field to the Amp table to indicate whether the time contained in Amp.datetime is exact or just an approximation. 3.- Allow NULL values in the Amp.datetime field and make use of the existing fields 'wstart' (Window start) and 'duration' to represent respectively the approximate time and its associated error. For every amplitudes that have an exact time, we would have: Amp.datetime = Amp.wstart Amp.duration = 0 For every amplitudes that have an approximate time, we would have: Amp.datetime = NULL Amp.wstart = Amp.duration =