Rainfall patterns will always amaze me.  Here is the CoCoRaHS map of Davis from this morning (i.e. data from 7am 11/17 to 7am 11/18).  The consistency of the rainfall gradient you see in that map suggests to me that this is truly indicative of a nearly doubling of received rainfall from the northwest to the southeast of town, rather than just measurement uncertainty.  The greatest distance between any two points on the map is probably about 2 miles.  How, then, does one forecast precipitation for Davis if the possible real, physical range of outcomes is a factor of 2?  Are our models considered a success if they get one of those values correct?  Or the mean?  Or, some scaled mean?  I don’t know.