Aggregation Series Slots in RiverWare 6.3
Additional Analysis Notes / Phil Weinstein / 5-25-2012

  1. We may want to constrain the possible options, especially focusing on the current "annual aggregation" requirements AND on uses of aggregation which we know to have already been used.  For example, we know that start-months for annual aggregations are needed, and that a 6am start-time for daily aggregations have been used with the SCT (by TVA).  But we might want to refrain from presenting a "start day" for monthly and annual aggregations (even though those are logically coherent features).

  2. In regard to the alternate start times for periods (e.g. "water year" and "water day" mentioned above), instead of having an independently settable start time for each Aggregation Series Slot, it may be better to support only a choice, on a per Aggregation Series Slot basis, between (1) standard timestep periods, and (2) one globally configured special timestep period (for each aggregation scope).  The semantics of having many different period definitions among, for example, all of the annual Aggregation Series Slots in the model, may cause unnecessary confusion.  Possibly having just a single unified alternate "water year" (other than the calendar year) would be preferred.  (Doing this is actually most analogous to the RDF Annualizer's function).  See this illustration:

  3. We may want to take on a resolution to the dysfunctional "weekly timesteps", if weekly aggregations are to be supported.  There doesn't seem to be a way in which weekly aggregation series can responsibly be presented without this fix.  A model-wide definition for a week-start-day would solve the problem.  (It's currently, effectively hard coded to Wednesday, i.e. midnight Tuesday).  
The image below illustrates several ideas -- and I'm going to represent a simplification of this.  (For example, I think we can drop the "start day" for both monthly and annual aggregations).  Also, the presentation needs some work.  So, this is maybe a little overwhelming (maybe close one eye) ... this is just to work out some ideas:

Single Panel:
Separate Tabs:

I'm still working on a more organized presentation of various options.

Here is yesterday's document, also still in progress:
http://cadswes2.colorado.edu/~philw/2012/AggregationSeriesSlots/InitialAnalysis.html

--- (end) ---