Intermezzo – Weather Station Latitudes

The previous post discussed the boom and the bust of the number of measurements available in the public NOAA temperature dataset. We will now try to find out where the measurements were made, and how the distribution of the weather stations has changed throughout the years.

As explained before, each line in the dataset contains the temperatures for the twelve months in a given year and for one weather station:

1016035500001984  125  104  125  150-9999  205  248  237  223  188  173-9999

In the example above, the year is 1984 and the weather station identifier is 10160355000. The first three digits of the identifier specify the country code; in this case 101 which stands for Algeria (from That country code only gives a rough idea of where the station is located; more precision is needed.

Fortunately, NOAA also provides the file v2.temperature.inv, which contains the metadata for all weather stations in the temperature dataset. In this file there is a line for station 10160355000:

10160355000 SKIKDA 36.93 6.95 7 18U 107HIxxCO 1x-9WARM DECIDUOUS  C

The station SKIKDA is located at 36.93ºN and 6.95ºE. With this information, the measurements could be placed on a global map. To show the evolution of the number of measurements over time by location on a map, a impractical graph with four dimensions would be required. By leaving out the longitude, and plotting the year on the X axis, the latitude on the Y axis and the number of measurements as the density, the following scatter plot for the dataset v2.mean emerges:

Number of stations by year and latitude

The first measurements in the beginning of the 18th century all come from Berlin at 52ºN. Over the years the range of represented latitudes gradually expands to cover both hemispheres – the first measurements from the southern hemisphere appear around 1840 – the tropics and the poles. As noted in the previous post, the golden years are 1950-1990. From 1990 onwards the number of measurements declines back to 19th century levels, but the range of latitudes remains wide.

Another thing that is obvious from eyeballing the plot is that the bulk of measurements has steadily drifted from higher to lower latitudes in both hemispheres. In the northern hemisphere the measurements started with a single station at 52ºN in 1701, whereas the center of gravity now lies at 40ºN.


About this entry