Introduction
About a year ago I started to write a descriptive post on the USHCN station at New Meadows, Idaho*. I got interested in the station because I used temperature data from it in assessments and reports with which I was involved during my time as a Fishery Biologist on the Payette National Forest. When I first started blogging, after I had retired, I posted an assessment of how the NOAA homogenization adjustments affected the data record. Despite this analysis, some of my interactions on Twitter indicated that it may not really be common knowledge that the data NOAA and NASA post as "official" are, in fact, adjusted. Since then, I think most people realize the official data are adjusted and the discussion has turned to how the adjustments, on average, reduce the apparent rate of warming; however, I have not found this to be clearly true here in Idaho (see here and here). Since the raw data from the New Meadows station clearly showed warming contrary to the adjusted record, I also did an analysis comparing the station's data trend with local RAWS data; their trends seemed mostly similar to the raw data at New Meadows.
During this process, I had conversations with folks involved with both NASA GISS and Berkeley Earth about the site, and was calmly assured that the adjustments were appropriate and that the temperatures were increasing despite the adjustments. At one point I was pointed to an analysis of USCRN data to document the regional trend and told that my RAWS analysis of regional sites was based on too short a record; that disappointed me because the USCRN data record was actually shorter! Then I heard some stuff about station metadata and station moves that didn't answer any questions but led me to looking into the validity of the metadata in relation to data analyses. Here's the one that intrigued me the most:
This chart shows two significant things: (1) station temperatures trend downward relative to their regional model (all of Idaho, I think) and (2) two station moves may have occurred. It was actually the latter that got me most interested and led to this post.
Station Locations
NOAA Metadata
One well-known climate expert with whom I had discussions referred me to the station's metadata when I was asking why the data adjustments were appropriate. I told him that the metadata were wrong, including the fact that the two station relocations that are shown in the above figure was probably inaccurate and surely incomplete; I was summarily accused of relying on "anecdotal" information and directed to the station metadata page, now duplicated (sort of) here. These metadata pages are really confusing and inconsistent, which I don't find strange because data collection and management standards have changed and tools for documenting locations have improved considerably in recent decades. But neither of these pages is very accurate and there has apparently been little quality control over the information posted.
The station has always (at least as long as anyone in the local area seems to remember) been associated with the USFS Ranger Station (note that the official site name is "NEW MEADOWS RANGER STATION, ID US" with ID "GHCND:USC00106388"). The monitoring equipment have probably been romed around when at any individual station (the metadat indicate this) but the ranger station itslf has been moved on occasion. In my view, two significant points regarding documented relocations are relevant: (1) it was almost certainly never at the open field location south of town labeled "4" from this page:
which looks like this in real life (the rural setting does not preclude it having been here, but no Ranger Station has been in recent memory):
and (2) it has been moved since 1958 contrary to this relocation history on this page:
There are several other issues, such as written site descriptions not matching Lat-Long positions and some apparent situations where the descriptive information may not have changed with station relocations, but the above are the most serious in my opinion. This confusion, along with my experiential knowledge being dismissed as "anaecdotal," led me to try and fill in some of the knowledge gaps by personal investigation.
What I Knew and What I Discovered
Late 20th Century Station
I began my tenure on the Payette National Forest in 1994. At that time the Ranger Station was located at the spot marked "2" on this Image**:
The ranger station was in this building, which was leased by the Forest and is east of downtown New Meadows:
I did not work at this station, but I visited it on many occasions. I'm not certain, but I believe there would have been a Stevenson Screen at the northeast corner of this property; I don't recall it, but historic imagery viewable on Google(TM) Earth show structures there and conversations with the Payette National Forest Archaeologist/Historian indicate he saw them east of the station and north of US Highway 55. It should also be noted that the building had a natural (dirt) surface parking lot., not a paved one and the area where I think the thermometer was would have been grassier or weedier because and adjacent to an irrigated meadow.
Mid-20th Century.
Mid-century, from at least the 1920s, the station was located in downtown New Meadows at the site identified as "1" in the aerial view above. This location is downtown, about one-quarter mile northeast of the Post Office (identified as "P" in the image). It was in this general area associated with the Ranger Station and his quarters*** when those were in town:
The New Meadows Post Office is the reference point for most of the descriptions of station location as here:
The old Ranger Station and the Ranger's house are now businesses in New Meadows, but a few Forest Service buildings like these one still remain:
What I couldn't determine was exactly when the station was put there (1920s will do for now) or where it was before that (I expect it may have been in Meadows [colloquially called "Old Meadows"] east of the highway junction in New Meadows (in this area, towns were occasionally relocated to affiliate themselves with important infrastructure developments like highways). Since the temperature record goes back to the 1800s, it almost certainly was somewhere else, most likely near US 55 and it is known that there was an Idaho National Forest Ranger Station somewhere in the area by 1908 (the Idaho and Weiser National Forests were consolidated in 1944 to form what is now the Payette National Forest). The USDA agency we know know as the U.S. Forest Service was created in 1905, but the USHCN station predates that.
Twenty-First Century Location
In about 2000, the Forest had a brand-new Ranger Station built west of town and just south of Highway 95 at the site labeled "3" in the aerial view above. It's a modern, uninteresting building but it does have two floors and probably the only elevator in town. The USHCN station instruments (an MMTS) are placed adjacent to the parking lot and some heat exchangers just west of the building:
Siting seems to violate NOAA standards, but I haven't taken precision measurements; the picture speaks for itself.
Temperature Record
I'm not going to spend much effort going back into the details of how the temperatures have been adjusted at this site because I already did that in previous posts. For this post, I've created a chart of raw and adjusted anomalies (not being careful to pick any particular baseline period, which I'm sure sets me up for some criticism, but, hey the professionals pick what they like so I may as well):
One obvious question appears immediately: With the station moved successively from a likely shaded spot (I can't confirm that; the trees are mature and probably old enough to have shaded the Ranger Station, but maybe not) to a rural but less shaded one, to a completely unshaded one near a parking lot and air conditioning units, why (a) did only the records before moving to that site need aggressive adjustment, and (b) why did all the older locations require cooling?
As indicated above, I have compared this site to nearby RAWS sites that are (I think) unadjusted and in relatively undeveloped locations. As also mentioned above, I asked a well known climate professional about this comparison who suggested that the RAWS record as a regionally coherent data set was too short and I should consider using USCRN; interestingly, as mentioned above, the RAWS set was actually over a longer term than USCRN. As for the much stated comparability of temperatures over wide geographic ranges, I offer this comparison of a RAWS site and a USCRN site about 100 or so miles away:
Here, I used maximum temperatures and some persons I've chatted with would rather see averages; using maximum does increase the disparity, but it's also closer to measured values. The reader is free to draw whatever conclusion he or she chooses.
A final point involves the direction of the trends. We hear regularly about how adjustments reduce the warming trend in temperature rather than increase it. That is not the case here using either USHCN estimates (upper) or the supposedly independent BEST estimates (lower):
Wrap-Up
The most important take home message for me, as a former data manager and analyst, is that almost nothing about the metadata is correct yet changes to the raw data are brushed of as corrections required to account for site specific changes. I've had many people tell me these are "known errors," which hardly seems to be the case. I concede that it is known that the USHCN system was not set up for global temoeratture monitoring (as used now) and that non-climatic factors undoubtedly exist in the database; however, I hardly think that justifies making up data when none of the specifics are known and no validation for accuracy is performed.
The metadata still seem to say that the station is northeast of the Post Office and it doesn't even indentify either of the 20th century station locations. It's easy to say that, on average, adjustments of USHCN records make an average of the records from those stations more accurate, but how do those adjustments when applied to individual stations affect the accuracy of the local record? I can grasp the sense of the global adjustment and have no problem with constructing an index for estimating global change, but I would expect that the results are thoroughly validated (yes, I know there are studies that affirm validation for the general index, and I am not questioning the premise); but what do we do locally? It's of interest to me, since I live here, to know what is actually happening here, not what is modeled to be happening here. My investigation suggests the adjustments may inappropriate locally. I might change my mind with specific justifications as to what, exactly, the adjustments are correcting and if any effort were made to correct the metadata and verify the necessity of the adjustments and validate their accuracy.
___________
* The CDIAC map-interface referenced in previous posts is no longer available; data should be available here via NOAA. I have not downloaded any data through this interface to date because of its inconvenience compared to the previous interface.
** The location marked "2" is the open field location mentioned above in the metadata discussion.
*** I spent quite a bit of time researching this. The New Meadows Postmaster asked several long-time residents what they knew, and I had conversations with the retired Payette National Forest Archaeologist/Historian. This is where the Ranger Station/Ranger house was before it moved east of town.
