In the debate over CO2-induced global warming, projected impacts on various weather and climate-related phenomena can only be adjudicated with observed data. Even before the specter of dreaded global warming arose, scientists studied historical databases looking for secular changes or stability. With the advent of general circulation climate models, using historical data, scientists can determine whether any observed changes are consistent with the predictions of these models as atmospheric carbon dioxide increases. An example of the pitfalls in such work was recently presented by Rahmat et al. (2015), who set out to analyze precipitation trends over the past century at five locations in Victoria, Australia. More specifically, the authors subjected each data set to a series of statistical tests to “analyze the temporal changes in historic rainfall variability at a given location and to gain insight into the importance of the length of data record” on the outcome of those tests. And what did their analyses reveal?
When examining the rainfall data over the period 1949-2011 it was found that all series had a decreasing trend (toward less rainfall), though the trends were significant for only two of the five stations. Such negative trends, however, were reversed to positive in three of the five stations when the trend analyses were expanded over a longer time domain that encompassed the whole of the 20th century (1900-2011 for four stations and 1909-2011 for the fifth one). In addition, the two stations with statistically significant negative trends during the shorter time period were also affected by the longer analysis. Though their trends remained negative, they were no longer statistically significant when calculated over the expanded 112 years of analysis. In summation, in the expanded analysis the “annual rainfall time series showed no significant trends for any of the five stations.”
In light of the above findings, Rahmat et al. write that “conclusions drawn from this paper point to the importance of selecting the time series data length in identifying trends and abrupt changes,” adding that due to climate variability, “trend testing results might be biased and strongly dependent on the data period selected.” Indeed they can be; and this analysis shows the absolute importance of evaluating climate model projections using data sets that have been in existence for sufficiently long periods of time (century-long or more) that are capable of capturing the variability of climate that occurs naturally. And when such data sets are used, as in the case of the study examined here, it appears that the modern rise in CO2 has had no measurable impact on rainfall trends in Victoria, Australia.
Rahmat, S.N., Jayasuriya, N. and Bhuiyan, M.A. 2015. Precipitation trends in Victoria, Australia. Journal of Water and Climate Change 6: 278-287.