Considerable scientific attention has been focused on a measured increase in atmospheric CO2 and a suspected corresponding change in climate. Such a change in climate, if it occurred, might be expected to have a magnified effect on hydrologic time series and, indeed, projections have been made of major changes in water resources.
If the climatic changes are indeed magnified in hydrologic time series then, by detecting trends in such series, it should be possible to work backwards and identify the causative climatic change. This paper looks at two data sets: 1) long-term temperature, precipitation and streamflow data from sites across Canada and 2) long-term levels of large lakes in Africa and North America.
The study assumes that time series may be modelled by trend, periodic, autoregressive and random residual components. The trend component of a time series is generally associated with changes in the structure of the time series caused by cumulative natural or manmade phenomena. Periodicities in natural time series are usually due to astronomical cycles such as the earth's rotation around the sun. Autoregressive components reflect the tendency for an event to be dependent on the magnitude of the previous event(s), a memory effect.
The analyses of temperature, precipitation and streamflow data show some significant linear trends but no pattern is apparent. The analyses of longterm lake levels also identify linear trends but these are all explainable without invoking climate change due to greenhouse gases.