Perth 1940 Max Min Daily Temps 16


Previous posts have introduced the work that Chris Gillham is doing in spot auditing the accuracy of the Bureau of Meteorology’s temperature records. He has now re-recorded the daily max and min temperatures from one Australian weather station for one year, Perth 9034 in 1940, using original sources in The West Australian newspaper.

Below is an initial look at the historic data (in red) compared to the BoM’s “unadjusted” or “raw” records (grey) for the station.

Its fairly clear that there are a lot of errors. The minimum temperatures, however, are shockers. Each of the red lines seen on the lower series above is an error in the daily minimum — mostly down.

Mean of the max differences = +0.20C
Mean of the min differences = -1.18C
Average max all differences = +0.04C
Average min all differences = -0.33C

While the average error of the max temperatures is up 0.2C, the average magnitude of the errors in the min temperatures is a whopping 1.18C! Over the whole year that changes the annual minimum temperature by -0.33C.

The diurnal range is increased by an average of 0.4C. While these errors are only in one year in one station, it is noteworthy that the magnitude of these errors is similar to the change in the diurnal range attributed to global warming.

The data file is here – perth-1940-actual-raw. You need to open it in excel and save as a CVS file.

The code below should run on the datafile.

P1940=ts(read.csv("perth-1940-actual-raw.csv"),start=1940,freq=365)
l=2
plot(P1940[,3],col=2,ylim=c(0,45),main="Perth Regional Office 9034",ylab="Temperature C",lwd=l)
lines(P1940[,4],col="gray",lwd=l)
lines(P1940[,7],col=2,lwd=l)
lines(P1940[,8],col="gray",lwd=l)
maxErrs=P1940[P1940[,3]!=P1940[,4],]
print(mean(maxErrs[,4]-maxErrs[,3]))
minErrs=P1940[P1940[,7]!=P1940[,8],]
print(mean(minErrs[,8]-minErrs[,7]))
print(mean(P1940[,4]-P1940[,3]))
print(mean(P1940[,8]-P1940[,7]))