[Previous] | [Session 41] | [Next]
N. S. Smith, F. Raulin (LISA, Univ. Paris 12)
In any scientific experiment, rigourous evaluation of potential uncertainty is crucial. Although this has not previously been emphasised, it is equally true for modelling studies if we are to place confidence in the model results. Errors arise from two main sources ; from the basic hypotheses of the model and from the errors that are present in the parameters used (e.g. rate constants, quantum yields). As the former source is difficult to evaluate, we have concentrated solely on the latter. Photochemical models of the giant planets and their satellites are particularly susceptible to this source of error as the low-temperature photochemistry of methane and its derivatives is poorly constrained by laboratory evidence. All parameters in photochemical models are interdependent and the system as a whole is strongly non-linear. Thus traditional sensitivity studies, that simply vary each parameter in turn, do not indicate the overall uncertainty in the model's results. For this reason, we have adopted a Monte Carlo approach, in which the photochemical model is run a large number of times, and on each run the values of all parameters are varied randomly, according to a pre-defined probability distribution. Statistically significant results are obtained after 5000 runs. Unfortunately, the large number of runs required limits us to 0-D box modelling, but this has been successfully applied to the study of laboratory experiments concerning hydrocarbon photochemistry. Our results show that the uncertainties in these simple 0-D simulations are very significant. Presumably, the errors would be even greater when extrapolated to a 1-D model of Titan's atmosphere. Techniques are being developed to pinpoint the photochemical parameters that are responsible for inducing the largest errors. In this way, it is hoped that key laboratory measurements can be suggested from which the more accurate data would reduce the uncertainties of the models. This shows that we need to reform the way we think of, and use, current photochemical models to understand the outer solar system.