The mysterious sudden jump in Melbourne temperatures in 1996 with an instrument change

Here’s a strange change. After 160 years of fairly constant maximum temperatures, the raw Melbourne records take a sudden step up by 0.7 ⁰C in 1996. Coincidentally (or not) that is the same year that the automatic gauge was installed. The new electronic equipment is much more responsive to short peaks and dips compared to thermometers. Could the step up be due to the better resolution? It’s by no means definitive — these are yearly averages, not monthly, and it may be a real climate shift and not due to the equipment. The obvious question is whether this sort of jump occurs in other stations where AWS (automatic weather stations) were installed. That would have profound implications if it did, but surely it would have been noticed already? Melbourne is known for having “four seasons in one day”, so perhaps there is a small effect in most places, but the switchable peaks of of Melbourne summers make a larger difference. In any case, thanks to Tom Quirk (and Bill Johnston)  we have another puzzle in need of an answer. These AWS’s were installed all over Australia in the late nineties. If there was some effect, then there would be a lot of artificial small step ups as better equipment started to detect faster, shorter peaks. The ACORN adjustments make no corrections to the max record in 1996 but there is an adjustment made to the minima then. Hmm. – Jo

Twenty-first century Melbourne temperatures

Guest Post by Tom Quirk

During the 1990s the Bureau of Meteorology (BOM) started to install automatic weather stations. The minimum and maximum temperatures along with temperature readings every 30 minutes were recorded electronically. The weather stations used thermistors rather than mercury or alcohol thermometers for temperature measurements. Thermistors are electrical resistors that are very sensitive to temperature changes.

The Melbourne annual maximum temperature readings are shown in Figure 1. There is an unusual feature in the record as shown occurring in 1996. It is a break with a difference of 0.7 ⁰C. The two straight lines are a best fit to the measurements and the difference is calculated from the lines at 1996.

Figure 1: Annual average raw maximum temperatures for the Melbourne Regional Office. The straight lines represent the best fit for two lines.

The ACORN-SAT adjustment record (Figure 2) shows only an increase of 0.41 ⁰C to the maximum temperature record starting at 1 Jan 1990. This adjustment is explained as “statistical”. Interestingly the adjustment record shows a break in 1996 for the minimum temperature record.

Figure 2: ACORN-SAT adjustments to the raw Melbourne temperature measurements as listed by BOM.

There is reason to believe that the BOM started using the automatic weather station readings in 1996 rather than the thermometer measurements.  If the weather station thermistors have a faster response time to temperature changes then short term increases may be recorded. This would be particularly the case for summer in Melbourne where hot north winds and cool changes produce rapid temperature variations and the Melbourne Regional Office site is shielded in part by large buildings to the south but more open to the north with the gardens surrounding the Exhibition Buildings.

There is evidence of this faster response time with a larger temperature break in summer than in winter. Summer is the months of December, January and February and winter is June, July and August. The maximum temperatures are shown in Figure 3. Simple temperature differences before and after 1996 are given in Table 1 and show the increased difference in the summer.

Figure 3: Summer (Dec, Jan, Feb) and winter (Jun, Jul, Aug) average raw maximum temperatures for the Melbourne Regional Office. The straight lines represent the best fit for two lines with a break in 1999

Table 1: Average temperatures before and after 1996

Conclusion

The break in the maximum temperatures in Melbourne in 1996 has not been detected in the BOM ACORN-SAT analysis. The break may be due to the shorter response time of thermistors compared to mercury thermometers. This would put the reason for the break on a better footing than a “statistical” explanation…

It would be interesting to know if other BOM sites have the same performance. But perhaps it is due to the peculiarities of the Melbourne location of the weather station. This follows Mark Twains sensible statement that climate is what you want and weather is what you get.