More on Central Idaho USHCN Data

My first post looked at temperature data from the U.S. Historical Climate Network (USHCN) site at New Meadows, Idaho. That analysis was a simple look at how, when, and by how much data had been estimated or adjusted. I picked that particular site simply because I was familiar with it and had used data from it in analyses I performed as a Fisheries Biologist on the Payette National Forest (PAF). This post adds to that analysis with a look at four other nearby sites, two to the north and two to the south, and the relationships among them.

These are the five sites investigated: New Meadows (106388), Cambridge (101408), Payette (106891), and Moscow (106152). The four additional sites were selected because they are the closest sites to New Meadows and are on an approximate North-South axis. Idaho has a temperature and precipitation gradient that runs more or less diagonally from northwest to southeast, and this arrangement serves to minimize the east-west influence. Elevations of these four sites and distances* from New Meadows are shown in this table:

New Meadows has the highest elevation of these five USHCN stations at 3,870 feet. West-central Idaho has high relief and elevation is certainly a factor, but no nearby USHCN sites are at the moderate elevation of New Meadows.

Firstly, I wanted to see whether, how much, and in which directions temperature adjustments were made. The data manipulations are summarized in this table:

where: Total is the number of "official" (i.e., SHAP adjusted) mean annual temperature records; Raw is the number of raw (unadjusted) annual mean temperature records; # is the number of records with non-zero adjustments; and % is the percentage of records adjusted. Clearly the vast majority of records have been adjusted in some fashion.

When looking at the adjustments, it was clear that the magnitudes (and occasionally the direction) of adjustments changed with time, mostly becoming lower in recent decades (as I showed in a previous post). This suggested three periods where adjustments changed in magnitude: Early (before 1910), Middle (1910-1949), and Late (1950 and beyond):

 Only one station doesn't vary much among periods, Payette. Oddly, this is one of the two (the other is Lewiston) where I might expect the largest increase in the Urban Heat Island (UHI) effect as it's in the growth area surrounding Boise.

I don't really know how the adjustments and estimations are made. The online USHCN documentation suggests that estimates involve using data from nearby, highly correlated stations. I looked at the correlations of the New Meadows data, where about one-third of the data were estimated, with these four stations where they had raw data for the same years, and I can't see how that makes sense here with these relationships:

Clearly, raw data from none of the stations are well correlated with the New Meadows raw data.

So data are mostly adjusted, mostly downward but decreasing in magnitude over time, and none of the sites can be considered highly correlated with the New Meadows station. Curiouser and curiouser, it would seem.
___________
* Elevations were obtained from the USHCN web interface page and distances acquired using MapQuest.