基于四维变分资料同化的核事故源项反演
Source inversion in nuclear accidents based on 4D variational data assimilation

福岛(Fukushima)核事故后,源项反演成为一种重要的核事故放射性释放源项定量计算分析方法。该文引入四维变分(4DVAR)资料同化法,结合中尺度大气扩散模型,提出一种针对核电厂事故的放射性释放源项反演方法。该方法利用核电厂周围监测数据,使用伴随方法迭代计算四维变分代价函数梯度,得到对释放源项的最佳估计。该方法考虑了完整时间序列上的放射性传输过程,对释放源项的估计结果为全局最优。风洞实验验证结果表明: 源项估计的相对误差为20%左右。
Abstract： Source inversion is an important way to quantitatively analyze the radioactive release source in nuclear accidents such as the Fukushima nuclear accident. This study presents an inversion method for radioactive release source estimates in nuclear accidents that couples a four-dimensional variational (4DVAR) data assimilation technique with a mesoscale atmospheric dispersion model. The method is formulated as an optimization problem between the real observation data and the background field. Environmental radiation monitoring data around the nuclear power station is used to calculate the gradient for cost function by integrating the adjoint to calculate the optimal source estimate. The advantage of this method is that the radionuclide transport is included in every time step in the data assimilation and the result is the global optimum over the whole assimilation period. The method was verified using the data of a wind tunnel experiment. The results demonstrate that source estimate errors are approximately 20%.