Cumulative values in erosivity calculation

[SethCallewaert]

I saw that there is a difference of inclusion in the cumulative columns yielded from the compute_erosivity function ('all_event_rain_cum' vs 'erosivity_cum'. The former does not include the values of the calculated storm, while the latter does. I do not have strong preference for one specific way of showing the cumulative values, however it should be consistent for both columns.

It also seems that there the all_event_rain does not add up with the sum of the event_rain_cum, maybe because not all rainfall is counted into the events (short rainfall or < rainfall than threshold). See attachments.

KMI_6438_2013.txt

[Sachagobeyn]

@SethCallewaert : from what I can tell from the code (https://github.com/watem-sedem/rfactor/blob/master/src/rfactor/rfactor.py#L398-L405)

• the datetime states the beginning of the event
• the event_rain_cum is the total amount fallen during that event
• max_30min_intensity is the maximum I30 over the event
  • event_energy: total energy of event.

the all_event_rain_cum is a sum of event_rain_cum (with a shift indeed, thus summing all rainfall untill end of event).

The reason why it does not seem to add is line here: the https://github.com/watem-sedem/rfactor/blob/master/src/rfactor/rfactor.py#L420 - event with a small amount of rain are removed. Can you check if this compliant with the workflow of GV?

[SethCallewaert]

My issue is in the columns 'all_event_rain_cum' and 'erosivity_cum'. If you look at the first line, the former is 0 and the latter is 2.08, which is strange as both should or take into account this event or not.

in line: https://github.com/watem-sedem/rfactor/blob/577efc0d2121e28dd996f332464b29f7cbb0e35a/src/rfactor/rfactor.py#L415 there seems to be a shift implemented, but what is the reason for this?

[Sachagobeyn]

@SethCallewaert:

• The shift is implemented so the current event is not included in the sum. This is relevant for situations where the start date of an event is later then the start date of the timeseries: the first index holds the cumulative rainfall before the first event, otherwise it includes the rainfall of non-erosive events before the first erosive event. See https://github.com/watem-sedem/rfactor/pull/102

Can you let us know what you prefer?