Linear stochastic systems
System identification (2nd ed.): theory for the user
System identification (2nd ed.): theory for the user
Sensitivity Analysis in Practice: A Guide to Assessing Scientific Models
Sensitivity Analysis in Practice: A Guide to Assessing Scientific Models
Adaptive wavelet neural network for prediction of hourly NOX and NO2 concentrations
WSC '04 Proceedings of the 36th conference on Winter simulation
| Hi-index | 0.00 |
In recent years computational models and statistical modelling have been increasingly coupled together. On the one side, statistics is a useful tool for computer simulation design and output analysis, namely importance, uncertainty and sensistivity analysis. On the other side simulation is becoming an important part of statistical modelling. Moreover, data assimilation based on statistical modelling allows us to manage both empirical data and computer outputs on a common ground. In this paper, we consider some techniques for decomposing and analysing the output uncertainty for environmental models with spatio-temporal components. This extends standard sensitivity analysis variance decompositions to the case of correlated inputs which is useful for assessing the adjusted anthropic impact on the environment through appropriate models. The model setup used is based on the hierarchical spatio-temporal approach, which allows to define various model components in a relatively easy way and to introduce both spatial and temporal correlation. Moreover, this gives a natural frame for discussing uncertainty decomposition in terms of model components and mapping capability. Two motivating applications are given. The first one involves daily data on particulate matters related to both a monitoring network and an emission inventory model coupled with a meteorological and chemical computer model. The second one is related to impact assessment of vehicle traffic on hourly carbon oxides in Turin, Italy.