tbohn
commented
7 years ago It's also worth noting that, because the time series is a linear interpolation between the daily VP values given by MTCLIM, the resulting hourly time series of VP (whether or not the dips occur) will not in general average out to the same daily average VP as that given by MTCLIM. It will be close, of course, but the biases will be larger on days that are local extrema of the daily VP time series. So, at least from the standpoint of consistency with the daily values, the linear interpolation should probably be replaced.
One idea could be (if you're willing to assume that VP always == saturation VP at sunrise, which I think is an assumption in MTCLIM anyway) to take the curve of hourly saturation VP computed from the disaggregated air temperature, select the period between successive sunrises, and scale (and offset) it appropriately to have the same daily average as MTCLIM and to equal saturation VP at sunrise. But of course the real determinant should be what fits hourly observations best.
See https://github.com/UW-Hydro/VIC/issues/667 for details.
Summary: The vapor pressure disaggregation shows briefs dips during certain hours of the day. As explained by @tbohn, these are intentional to ensure that the relative humidity does not exceed 100%. However the resulting time series looks "odd" (technical term). Because the behavior was intentional, the decision was made to replicate the behavior in MetSim 1.0. However, we may want to evaluate whether this is the desired behavior.