Выпуск: 2 , Том: 152 , Год издания: 2003

The formation temperature below the sea or lake floor is frequently measured with a sensor probe that is inserted into the unconsolidated sediment, and the thermal conductivity of sediment is measured in situ with an independent heating experiment. Friction during the insertion raises the temperature of a probe. This paper presents the method and the test results of using the cooling history of friction-heated probes for the determination of equilibrium formation temperature and thermal conductivity. The inverse modelling (IM) for the desired parameters is formulated together with the finite-element simulation of the cooling history. The starting parameter values for IM are generated with a genetic algorithm (GA) that searches for them in the likely parameter ranges. The IM is constrained in an objective function by a commonly used asymptotic relation between the declining temperature and the inverse time. In addition to satisfying the root-mean-square misfit, the results are assessed by the equality between the model conductivity and the conductivity obtained through the asymptotic relation and by how well the models can predict the cooling behaviour at time beyond the record period for IM. The results of testing synthetic data with random noise up to ±0.005 K and twenty 8-11 min long data sets from four sites in Lake Baikal indicate that the desired parameters can be determined from the first 2.5 min of the cooling records. By comparison with independent measurements or through repeated GA-IM, equilibrium temperatures can be consistently determined to within 0.002 K and thermal properties typically to ±5 per cent. The method has also been successfully applied to seven sets of 5 min long data sets at one marine heat flow station.