Duration of Deccan trap volcanism: a statistical approach

Abstract

We examine the completeness of the magnetostratigraphic record in sequences of lava flows where few reversals have been recorded. A case of particular interest is the Deccan traps in India, which appear to exhibit on the order of 50 flows, with a three chron (NRN) polarity sequence and a strong bias to reversed polarity (80%). K/Ar ages only weakly constrain the duration of volcanism between 50 and 70 Myr. Simple statistical experiments in the form of random sampling of the reversal time scale in that time window show that volcanism is unlikely to have lasted more than 3 Myr. The range of acceptable chrons can be reduced with paleontological arguments to the 29–32 sequence. We have tested the probability of losing short chrons in that time window and as a result observing only a NRN sequence, when actually sampling 4 or 5 chrons. This probability is around 35% for 50 flows and 20% for 75 flows. Only the shortest chrons 30R and 32.1N can be missed: as a consequence possible correlations of the Deccan main reversed sequence are restricted to 29R and 31R (the very short chron 30R being unlikely from geological considerations). Recent paleontological data actually seem to restrict the correlation to 29R. In all cases, the observed polarity bias can be reproduced only if volcanism lasted less than 1 Myr.