NOAA’s National Centers for Environmental Information (NCEI) publishes a temperature database for 7280 worldwide meteorological stations. It is called the Global Historical Climatology Network (GHCN.) The database consists of six files, three (maximum, minimum, and average) for ‘Unadjusted’ and three for ‘Adjusted’ values. The Unadjusted was formerly called ‘Raw’ data. For this short essay only the average-value data set is under consideration.

The Unadjusted is ‘as received’ data from agencies/centers from around the world that collect the respective Station data. For the United States there are six collection centers known as NOAA’s the Regional Climate Centers. Supposedly the Unadjusted data are ‘as measured’ values, but the NOAA-NCEI advises on its GHNC web page “…it is entirely possible that the source of these data (generally National Meteorological Services) may have made adjustments to these data prior to their inclusion within the GHCNM.” NOAA-NCEI then makes its own adjustments, primarily a lowering of the temperature values of land stations in the earlier part of the temperature record. The Adjusted data are used to determine global temperature trends, the results of which are the basis of the claim that global temperature is rising at a significant and concerning rate, the cause of which is primarily attributed to human activities. Numerous scientists, engineers, and others have challenged this, asserting that NOAA-NCEI’s adjustments cause a temperature increase that is artificial. But this “adjustment” issue aside, there is at least three more fundamental issues with the GHNC database. They may be found by inspection of the GHNC Unadjusted average-temperature database.

The first issue pertains to NOAA-NCEI’s categorization of the stations’ locations. For the categorization of the stations used for the V2 database, the same as used for the V3 database, NOAA-NCEI categorized 7220 of the 7280 Stations according to whether they are in Urban, Suburban, and Rural locations. The agency then further divided the stations in each of the three locations into whether or not they were at airports. The first issue is that if the longitude and latitude values attributed to each of the stations are correct, then some of the stations said to be at Rural locations are not at what should be termed “rural.” Such errors may well be human, so perhaps they do not warrant too much complaint if the number of them are not excessive. Then again, rooting out errors and correcting them is a good engineering/scientific practice.

The second issue is more significant and pertains to stations involving airports. At least some of these stations began reporting data decades earlier, when it is highly unlikely there was an airport at that location. Thus, at dates earlier than when the V2 database was created, those stations are incorrectly biasing the data and should be in the ‘not at an airport’ category. Another aspect this issue is that for the Rural stations, NOAA-NCEI designates as being at airports the yearly average temperature jumps approximately 2 degrees C at about the year 1950. Some, if not all, of this discontinuity is due to the stations reporting before 1950 and those reporting after 1950 are entirely different stations at entirely different locations. This particular temperature discontinuity and the inaccurate categorization of stations at an earlier time being at airports when they were not carries over to data from all 7220 stations. The net effect is a temperature trend interpreted as an indication of a rise in global temperature, when instead the cause is documentation of station location that does not take into account when stations at airports actually began to be at airports. There should be some indication in the database when the host airport actually went into operation. Clearly, this adds to NOAA-NCEI’s report burdens, but as said earlier, rooting out errors and correcting them is a good engineering/scientific practice.

The third issue is perhaps far more significant than the first two. The durations stations report temperature varies from as little as 8 or 10 years to nearly, if not all, years for a timeframe, say 1900-2016. To exacerbate matters, some stations report data for only a few years in one portion of a timeframe and others, which are at entirely different locations, report for only a few years in much different portions of the timeframe. So for a particular timeframe station-reporting ‘pops in and out’ causing the averages to increase or decrease depending on the location. If a required reporting percentage of the timeframe is imposed, say 25, 50, or 75 %, then the temperature trend for the timeframe changes. For the cases of Urban and Rural locations, each at and not at an airport, a larger required reporting percentage results in a smaller value temperature trend for the timeframe. So the issue is that the important ‘quality-of-data’ factor is being overlooked in NOAA’s assessment of global temperature trends.

The question is whether or not NOAA is prepared, or even able, to address these three issues. Solving these alone will likely make for a significant change in the assessment of global temperatures. The reality may be that politics are much more significant issues. I have written a technical report on these issues. 

Edward R. Long is a retired NASA physicist who from time-to-time succumbs to the urge to inspect temperature and other databases.