An empirical thermal history of the Earth's upper mantle

Publisher Website
Google Scholar
Abstract
We have compiled petrological and geochemical data from 71 ophiolite suites and greenstone belts, which range in age from 15 to 3760 Ma. We have selected those rocks whose compositions indicate that they are either normal mid‐ocean ridge basalts (MORBs) or hotspot‐type MORBs. Then we used the data base to calculate the most primitive liquidus temperature for each rock suite. The results show that the liquidus temperature of the Phanerozoic ophiolites ranges from a low of 1212°C to a high of 1417°C. Using these data and two exponential curves bracketing the maximum and minimum temperatures versus time, we infer that the Phanerozoic suites had a mean liquidus temperature of 1272±7°C and a mean temperature range of 1218° to 1425°C. The liquidus temperatures of Archean MORBlike greenstones range from 1305° to 1576°C. Using these data and two exponential curves bracketing the maximum and minimum temperatures versus time, we infer that Archean melts at 2.8 Ga had a mean liquidus temperature of 1399±13°C and a temperature range from 1301° to 1533°C. Using two different methods, we show that the change in the mean liquidus temperature since the late Archean is from 96±13°C (from temperature ranges) to 127±20°C (from temperature means). When we convert these liquidus temperatures to potential temperature of the mantle, we find that the change in the mean upper mantle potential temperature since the late Archean is from 137±8°C (from temperature ranges) to 187±42°C (from temperature means). This change is less than that which was previously thought to have occurred. We compared the liquidus temperatures calculated from our data set with an independent data set from the modern day Pacific plate. The resulting histograms have the same shape and the same temperature range, showing that our method for calculating mantle temperatures from MORBlike rocks in ophiolite suites is valid. When our calculated liquidus temperatures for all time intervals are plotted in histograms, the resulting distributions are not bimodal, but skewed unimodal. That is, the distributions show a high‐T tail which results from the presence of hotspot magmas in the data set. The Archean temperature distribution is also skewed unimodal, and the high‐temperature Archean rocks, such as komatiites, plot in the hotspot area of the distribution. This strongly supports the contention that komatiites do not represent “normal” Archean mantle but rather were probably erupted by hotspots. Our data suggest that the relative proportion of hotspot magmas in oceanic lithosphere has remained nearly constant over geologic time.