In the event of a nuclear accident, radioactive materials may be released into the atmosphere, dispersing with winds and depositing on the ground. Inverse modelling can utilise observation data for backward tracing to the origins, and play an important role in consequence assessments of nuclear accidents. The model can estimate the nature and release of radioactive substances based on radiation monitoring and meteorological data, thereby assessing the severity of a nuclear accident and providing a scientific basis to optimise response measures.

The Observatory has developed an inverse modelling system comprising short-range and long-range modules. The short-range module is responsible for reconstructing the source term (i.e. composition and amount of radioactive substances released) while the long-range module mainly aims to estimate the potential location and timing of a release of radioactive substances. In the scenario of a known-location accident, the short-range module can utilise the radiation monitoring data in the vicinity of the accident, e.g., variations in environmental radiation levels or concentrations of detected radioactive substances, to compare with the dispersion simulation results based on different estimated source terms and determine the most appropriate source term estimation through statistical methods (Figure 1), in order to enhance the reliability of simulation data.

For the scenario of an accident with an unknown location, the long-range module can first perform backward trajectory analysis using radiation monitoring data to generate different possible trajectories. Based on the distribution of backward trajectories over different periods, possible regions of a radiological release and corresponding timing could be identified from the clustered areas of the trajectories (Figure 2). With the rough estimate ready, dispersion simulations over those regions can be performed. Correlations between simulation results and monitoring data can be studied to identify the most probable location and timing of the accident as well as the amount of its associated release. If prior knowledge of the distribution of nuclear or radiological facilities over the regions is available, the inverse process can be more specific to those facilities, reducing the time required for analysis.

The Observatory will streamline data processing of the inverse modelling system and further explore different simulation methods to enhance operational efficiency to better support emergency response and planning.