Flat to steep transition in subduction style

Downloaded from geology.gsapubs.org on February 16, 2015

Flat to steep transition in subduction style Rebecca^Drury } Walter H. F. Smith

Lamont-Doherty Earth Observatory, Palisades, New York 10964 NOAA, Silver Spring, Maryland 20910

ABSTRACT We use a thermal history of the mantle to calculate the evolution of oceanic crustal thickness over Earth history and use residual depth anomalies from the present-day Pacific to find the crustal thickness range (9.2-11.6 km) where the subduction style changes from mainly flat to mainly steep. We find that steep subduction was well developed by 2.5 Ga, which coincides with a major change in sedimentary rare-earth element patterns. Over 50% of all oceanic crust subducted steeply by 2.0-1.6 Ga, the same interval over which the average thickness of continental plates declined rapidly. Because producing thick plates requires many episodes of flat subduction, our model can explain why there are no known thick plates 2.5 Ga) to mostly higher K granites in the Phanerozoic (Condie, 1989). The low-K granites are a minor part of the Archean suites, which are primarily basaltic (Nisbet, 1987). The low-K granitic plus basaltic suites weather to form sediments with flat rare-earth element (REE) patterns, whereas the high-K granitic suites weather to form sediments with steeper R E E patterns and Eu depletion. Sedimentary R E E patterns show a major transition at ca. 2.5 Ga (Taylor and McLennan, 1981). Another change is from thick (>200-250 km) continental plates before 2.0-1.8 Ga to thinner ( O. V t(0

E

1400

•a

1300

3

O" 1100

1200 1000

2000 Age (Ma)

3000

1200

1300

1400

1500

Liquidus Temperature (°C)

4000

PACIFIC PLATE: 17 AGE GROUPS

Figure 1. Primitive liquidus temperatures (squares) for ail MORB-like suites vs. age. Because cooling temperatures are probably caused by declining radioactive heat production, exponential bounding curves approximate shape of terrestrial radiogenic heat production curve of Wasserburg et al. (1964). Figure after Abbott et al. (1994).

The relation between the residual depth anomaly and the crustal thickness is tc = (ra!218) + 7.1,

(1)

where tc is the crustal thickness in kilometres, and ra is the residual depth anomaly in metres (Abbott et al., 1994). 1200

RESULTS FROM TEMPERATURE CALCULATIONS The range of temperatures from the Phanerozoic MORB-like suites and the Pacific plate are 205 and 237 °C, respectively. Both distributions are skewed, unimodal distributions (Fig. 2). To simulate the variability associated with small groups of samples like our MORB-like suites, we divided the Pacific plate temperature data into 17 age groups. The standard deviation of the temperature ranges in these groups is ±30 °C, which implies that the temperature ranges of the Phanerozoic ophiolite and the Pacific plate data are not significantly different. We also found that the shape of the temperature distribution derived from Archean MORB-like suites was similar to the shape of the temperature distribution derived from the Pacific plate (Fig. 2). The slightly broader, lower peak results from adding together suites with an evolving temperature history. For example, adding together two temperature distributions with a mean offset of 30 °C reduces the peak height to 55% of its original value. The mean temperature range of our middle and late Archean MORB-like suites is 238 ± 29 °C, not significantly different from the temperature range of the two Phanerozoic data sets. The most important difference among the distributions is the uniform 30.5 °C offset between the liquidus temperatures derived from Phanerozoic ophiolites and the liquidus temperatures derived from the Pacific plate data. We wished to use our thermal history of the mantle to estimate the distribution of oceanic crustal thickness through time. A 30.5 °C offset causes an unacceptable error of —2.8 km in inferred crustal thickness. Therefore, we must decide which of the liquidus temperature distributions is most reliable. The enormous size of the Pacific plate database (15 000 samples as opposed to 36 Phanerozoic ophiolites) makes it less noisy and less subject to sampling biases. In contrast, the Phanerozoic ophiolite data have two factors that could produce slightly high temperatures. The first is that some Phanerozoic ophiolites formed in back-arc 938

1300

1400

1500

Liquidus Temperature (°C) 30-1

>2400 Ma SUITES O— «—

MAX (%) MEAN (%) MIN (%)

20-

>. o c

0) 3

O" 10w

1225

1325

1425

1525

1625

Liquidus Temperature (°C) Figure 2. Liquidus temperature distributions for three data sets. A: Phanerozoic MORB-like suites. B: Modern day Pacific plate data divided into 17 smaller groups. Note overall pattern of variability. C: Archean MORB-like suites. Groups A and C use error estimates to calculate minimum (MIN) and maximum (MAX) histograms.

settings (Nicolas, 1989) where the subducting slab adds water to the mantle. Because hydrous melting produces the same degree of partial melting at lower temperatures, calculated anhydrous primitive liquidus temperatures are too high. The second factor is that, as the surface area of a MORB-like ophiolite increases, it increases the chance of finding higher temperatures. We simulated this positive bias by finding the range among the mean temperatures in groups of 0.5 degree squares (deg 2 ) on the Pacific plate. The temperature bias ranged from +25 ± 11 °C for an ophiolite the size of Oman (4.0 deg 2 = groups of 8) to + 8 ± 6 °C for a small ophiolite (1.0 deg 2 = groups of 2). In contrast, the crustal thickness data from the Pacific GEOLOGY, October 1994

Downloaded from geology.gsapubs.org on February 16, 2015

120°E

300°E

23

E 40°N

21 • 19 ™

H STEEP • flat

17 m m m CO CO

15 0> 13 c o

10 km - cutoff

11

7 5 1 0

40°S Figure 3. Areas of steep subduction (squares) and flat subduction (crosses) iri Pacific. Note that most subduction is steep. Flat subduction makes up