Digital twin and scenario simulation: scope and limits of uncertainty
A digital twin oriented towards the management of critical infrastructure is not a static visualisation model, but a computing environment that replicates the dynamic behaviour of a physical system from continuous observational data. When used to simulate stress scenarios, such as flooding in an urban catchment, a mass evacuation or an extreme heat episode, the twin operates by integrating heterogeneous information layers: field sensor data, satellite Earth observation imagery, high-resolution meteorological models and demand parameters derived from statistical sources. The coherence between these layers directly conditions the quality of the results and, therefore, the utility of the simulation for the operational decision-maker. The execution of scenarios within the digital twin follows a sensitivity analysis logic: a set of input variables is systematically perturbed, such as the inflow rate in a water network or the spatial distribution of vulnerable population, in order to observe how the response of the modelled system varies. This process makes it possible to identify the nodes of greatest fragility and to prioritise interventions. However, the precision of the result is bounded by the quality of the underlying model, which incorporates simplifications of physical reality, and by the uncertainty inherent in the input data, whose propagation through the model can amplify local errors to produce significant deviations in the output. One of the most technically demanding aspects is the validation of the model against documented historical events. Without a rigorous phase of calibration and verification, the digital twin may reproduce plausible but not necessarily correct behaviours. Systematic biases in the model, those errors that are not distributed randomly but accumulate in a given direction, are especially critical in mission-critical decision contexts. The detection of these biases requires the availability of reference data series that are sufficiently dense and representative of the application domain, a condition that is not always met in urban environments with heterogeneous instrumentation or in regions with limited observational coverage. The honest communication of uncertainty limits to the institutional decision-maker is a technical responsibility that must not be subordinated to operational pressure. A simulation result must be accompanied by confidence intervals, an explicit description of the model assumptions and a qualitative assessment of the scenarios in which the model may degrade its performance. This transparency framework does not weaken the utility of the digital twin; rather, it allows the infrastructure operator or public administration to correctly calibrate the weight that should be assigned to the simulation within a broader decision-making process, one that includes expert judgement, established protocols and contextual considerations that no model can capture in their entirety.
NASSAT - Network Satellite Systems