NEW STRATEGIES FOR EXTRACTING DENUDATION CHRONOLOGIES FROM THERMOCHRONOLOGICAL DATA

Thermochronology is currently one of the main methods used to quantify rates of long term denudation. There are several stages to deriving a denudation chronology such as : appropriate quality data, a suitable predictive model (for the temperature dependence of annealing or diffusion), a robust technique for inferring the thermal history and a sound physically-based approach to converting the thermal history to an equivalent denudation chronology. Although each of these stages has variable degrees of uncertainty, one of the major limitations with current methods is that each sample is treated independently and as such common information is not explicitly exploited between samples.

Given the relatively coarse resolution of temperature from thermochronological data, we expect some spatial correlation between the inferred thermal histories. This may be difficult to characterise in many situations, depending on the samples’ geographical separation. More obvious is the implicit link between locally offset samples (obtained from vertical relief over 1-3 km). In this case, it is reasonable to assume that the sample currently at the highest elevation will always have been cooler than the sample at the lowest elevation (in the absence of faulting, fluids, etc). Incorporating this information explicitly into a modelling strategy yields physically consistent thermal history solutions, and also can provide a direct estimate of the palaeotemperature gradient within the sampled section as a function of time. Palaeoheat flow is obtained by combining this estimate with measurements of thermal conductivity on the rock types present in the sampled section.

However, the most important aspect of this approach is the improved resolution of the timing of enhanced cooling. This is because the time parameters in the thermal history are required to satisfy all the data simultaneously. This strategy can be applied to any type of data, and even provides a means to incorporate different data types into one model, in a meaningful way. Examples will be given to demonstrate the methodology and its application.