Data Adjustment: Inspection of West-Central Idaho Temperatures

This is not a post about Climate Change being a hoax, nor is it a denial of Anthropogenic Global Warming (AGW), catastrophic or otherwise. It is also not a specific criticism of data adjustments; as a former data manager and analyst, I recognize that adjustments and estimations are sometimes appropriate. The appropriateness is the question, which I won't be able to definitively answer; however, the following may help readers draw their own conclusions.

New Meadows, Idaho, is a small, rural town in west-central Idaho. At about 4000 feet above mean sea level, it sits in a mountain meadow on the upper reaches of the Little Salmon River. It is also home, on the grounds of the US Forest Service's New Meadows Ranger Station, to a weather station that is a component of the U.S. Historical Climatology Network (USHCN, station 106388). This site has a monitoring record of more than 100 years and is available here. Since the Forest Service consistently insists in its project assessments under the National Environmental Policy Act (NEPA) that AGW is occurring and must consider potential interactions with it's projects*, this should be an important resource; however, it's data are rarely used in this context, relying on more generalized warming forecasts instead.

In fact, I had used the USHCN data in some assessments I produced when I worked on the Payette National Forest (PAF), and the data never appeared to support any particularly significant warming. After I retired I downloaded the data because I wanted to use it in a commentary on a proposed Forest project. What I found was surprising: It seemed to me that there was less raw data than I remembered, though there was a complete sequence of adjusted data. I admit here though that that assessment was a "rush job" and I may have mistaken adjusted data for raw data. That's water under the bridge now, but a recent post by Christopher Booker about global temperatures actually being in decline for the past three decades and not increasing caught my attention because there is so much talk of government agencies falsifying data (e.g., here). So I decided I'd like to look at these data again.

So what do the data look like? I've plotted the SHAP** adjusted record here:

So there's considerable annual variability with a minor positive trend and approximately 2°F increase in 120 years. Not particularly worrisome as one would really expect temperatures to be increasing after the low points of the Little Ice Age and, more recently, the relatively cool 1960s and 1970s.

But how does this record compare to the raw data record? In short, not well. Here I've overlain the two data sets on the same axes:

Now there seems to be reduced variability and a slightly downward trend of about 1°F over the 120 years. Well, less than 120 years because you can see that data before about 1905 don't seem to exist. So here's a few statistics***:

Interesting. I don't know enough about the station history to know whether the changes were appropriate. I know the station moved about a half mile with a Ranger Station move in the early 2000s, which I wouldn't expect to matter much because of the rural setting. Equally clearly, prior to 1979, the data values were reduced in the adjustments much more than they were after that. (I've not shown it here, but the few raw data points available before 1910 were reduced by about -3.5°F). Visually, the adjustments over time look like this:

There may be more information available in the station metadata that I have not yet examined, so this analysis is hardly exhaustive.  But this is a very small, rural town probably not subject much to UHI effects. It seems odd that all measurements prior to 1979 were all missing, erroneous, or 1.8°F too high on average; that is, that all data needed to be estimated or adjusted downward. There are raw precipitation data for two years prior to the start of the raw temperature record, but one of those clearly seems wrong, suggesting problems with the station's data generally prior to 1906; but why recorded temperatures would have been 3.5°F too high is unknown to me.

I confess to being skeptical that anything out of the ordinary, climate-wise, is happening; simple investigations like this fuel such skepticism. I also discount prognostications of dangerous effects of AGW, if they're happening, as inherently unscientific. Speculation is appropriate in science for generating theories and testable hypotheses subject to falsification. In other words, science is about evidence and demonstrable facts. It does concern me, however, that the adjustments in this data set serve primarily to make the past colder relative to the present. 

It also concerns me that the PAF assumes climate change is creating warmer temperatures when local data may not support that assumption. The PAF manages land in a relatively small area in west-central Idaho. Local conditions, not just theoretical global changes, need to be assessed on their own merits for potential project effects. 

Update (01-30-2019): The use of the abbreviation "SHAP" was incorrect. The SHAP homogenization method was version 1 and was replaced by the "version 2 "pairwise" homogenization algorithm," the adjustment reported here.




__________
* Federal law does require consideration of climate change and greenhouse gases in NEPA analyses.
** Data are adjusted for time of observation (TOBS) and station history (SHAP), which presumably effectively incorporates TOBS (see here).
*** Adjustment values are for no-zero adjustments; there were 4 adjusments that produced no change, all since 2009.