How NCDC’s Datasets Compare to Each Other

Figure 1

Figure 2

On August 8, 2012, NCDC reported that July 2012 was the hottest month since recordkeeping began in 1895. It was 3.3 degrees (F) above the 20th-century average, which amounts to an absolute temperature of 77.6° F. NCDC developed the July temperature data from its key U.S. dataset called U.S. Historical Climatological Network v2 (USHCNv2). We use USHCN because it goes back to 1895 and provides the longest period of record.

But what about NCDC’s new U.S. Climate Reference Network (USCRN)? This network, only in place since 2004, is located in pristine locations and provides the highest quality climate observations. How did the temperature in this network compare to the longer USHCN record? Looking at just absolute temperature, USCRN stations average 75.6°F for July 2012. This looks cooler than USHCN, but the different results are a function of the technique used to calculate the average temperature, not a reflection of a true difference in temperature. So this is not an accurate comparison.

So how do we compare these two datasets correctly? To accurately compare national average temperature using USCRN and USHCN, it is necessary to use temperature departures from normal, also called the anomalies. This shows how much each station is warmer or cooler than its average temperature. When looking only at absolute temperatures, data are greatly influenced by topographic features such as oceans, lakes, rivers, mountains, and deserts. For example, a station at sea level can be 80°F while a nearby mountain station temperature is 60°F. However, both stations will typically have similar temperature anomalies and be similarly warmer or colder than average. The strong agreement of temperature anomalies over great distances regardless of topographic feature allows national temperature anomalies to be computed with fewer stations.

It is also critical to compare to the same time period. To compare these two networks the same reference period should be used for both networks. Using the same base period of 2006–2010, and the values for U.S. mean air temperature available from the NCDC monthly climate report, the 2006–2010 average is subtracted from both datasets. This yields mean air temperature departures that are very nearly identical. This can be seen graphically (Figure 1).

When looking at anomalies instead of absolute temperature and using the same base period, making it an apples-to-apples comparison, the USCRN and USHCN are in full agreement. This also confirms that July 2012 was the warmest July since 1895.

For more information see...
The July 2012 Monthly Climate Report
The Climate Reference Network Website
The U.S. Historical Climatology Network Website