option to short-cut pre-pended data requirement

[andywood]

andywood --
when the time is right, it would be helpful to consider a usability modification -- to remove the requirement for a 90-day prepend period. This period is used to calculate an effective annual precip to switch between an arid or humid condition in the solar radiation calculations, if I recall correctly (dep. if ann pcp > 8 cm or not). This is cumbersome/inefficient if one is using metsim in certain contexts, eg, to disaggregate 7-day ensemble predictions. I'd like to create an option allowing one to short-cut that analysis & requirement by inputting user-prescribed initial settings for ann pcp and any other prepend variable, which could then be overwritten for longer timeseries by the actual data, or not. This is just a proposal, since metsim is a shared tool -- others might want to handle it differently.

arbennett --
I think a feature working around/removing the need for the 90 day state file would be great, it is definitely an annoying usability point. The other variable computed using the state file is a "smoothed" daily temperature range, which is just a 30 day rolling average.

bart --
As long as we maintain exact restarts …. (or at least the ability for exact restarts)

[andywood]

the DTR requirement may be less critical, ie one could just substitute the DTR of the first N days and let the rolling window start to populate from there or be forward looking. The default no-prepend option could be the same for precip -- if it's a N-day sequence being disaggregated, where N<90, the wet/dry criterion is calculated from the precip in the available days looking forward. A 2ndary option would be that the user could override this with an input file (or scalar) containing the desired annual pcp, perhaps from climo. and the original option of prepending data would be untouched.