One of the highlights of the new 1981-2010 normals is that NCDC has employed some fairly standard statistical techniques to develop normals for stations with much less than the full (or nearly full) 30 year period. In part this has become practical with the great increase in computing power over the past 20 years rather than any magical breakthrough in statistical or climatic understanding. That said, here are plots of the 1981-2010 daily normal max and min temperatures for Keystone Ridge (1600' MSL) and Goldstream Valley Bottom (500' MSL, known to NCDC as Ester 5NE). There stations are about nine miles apart in the same drainage, so this is about as close to an apples to apples comparison as there is.
For normal highs, temperatures, Goldstream Valley Bottom runs colder than the ridge from late October until late February.
For normal min temperatures, the valley is always cooler than the ridge, though of course the differences are much larger during the snow cover season.
The other feature of both plots is that for the first time, NCDC has utilized the actual daily values in computing the daily normals. Duh, you say? In the past, NCDC has used a simple spline function that basically mapped a smooth curve of the monthly means to the daily values. As a result, daily normals were simple sine curves that mostly varied in how sharp the extremes were. With these new normals, much more structure in the daily normals. Notice the flat stretch in normal max temperatures during late November and early December at Keystone Ridge.