Inverse modeling of the ¹³⁷Cs source term of the Fukushima Dai-ichi Nuclear Power Plant accident constrained by a deposition map monitored by aircraft

The Fukushima Dai-ichi Nuclear Power Plant (FNPP1) accident subsequent to the great earthquake and tsunami of 11 March 2011 released large amount of radionuclides into the atmosphere. The discharged radionuclides were transported, dispersed, and deposited worldwide. To understand the spatiotemporal characteristics of atmospheric transport and deposition, various studies have been conducted with atmospheric dispersion models (ADMs) in regional (Morino et al., 2013, Morino et al., 2011, Park et al., 2013, Yasunari et al., 2011) and global (Christoudias and Lelieveld, 2013, Stohl et al., 2012, Takemura et al., 2011, Ten Hoeve and Jacobson, 2012) scales. In addition, a study comparing the models assesses the current status of and uncertainties associated with radionuclide simulations (Science Council of Japan (2014)).

ADMs include various processes (e.g., advection, dispersion, chemical reaction, and wet and dry deposition), and use the source term and meteorological fields as inputs. Although each process and input contains errors, one of the largest contributors to uncertainty is the source term because bias in the source term can directly affect atmospheric concentrations and deposition amounts. Inverse modeling, which optimizes the source term by integrating ADMs and observations, is one way to reduce uncertainty and obtain better simulation results. Several simple methods for source estimations studies have been performed for the FNPP1 accident with different ADMs and observational dataset. Researchers in the Japan Atomic Energy Agency (JAEA) performed source estimations with a regional model and data consisting of daily and monthly surface deposition amounts and air concentrations measured in Japan (Chino et al., 2011, Katata et al., 2012, Terada et al., 2012). More recently, Katata et al. (2015) revised the source terms estimated by Terada et al. (2012). Researchers in the Institut de Radioprotection et de Sûreté Nucléaire (IRSN) assessed emission peaks and the quantities with the facility events and ambient measurements (Korsakissok et al., 2013, Mathieu et al., 2012). Stohl et al. (2012) and Winiarek et al. (2012) performed inverse modeling studies on a global scale with atmospheric activity concentrations measured by a global International Monitoring System operated by the Comprehensive Nuclear Test Ban Treaty Organization. They also demonstrated that using a priori knowledge of the source term (an a priori source term or first guess) can compensate for the lack of observed data, and their inverse modeling successfully improved a priori source term. Saunier et al. (2013) estimated emissions of radionuclides with dose rate measurements. More recently, Winiarek et al. (2014) developed an inversion method using cumulated deposition and air activity concentration simultaneously. However, the range of emissions estimates was wide even in the inverse modeling results.

To refine estimates of ¹³⁷Cs emissions from the FNPP1 accident, we performed an inverse modeling study with a regional ADM, an a priori source term, and ¹³⁷Cs depositions measured by aircraft. The aircraft observation yielded a detailed map of ¹³⁷Cs deposition over the eastern part of Japan’s main island. More than 16,000 data points were used in the inverse modeling, much larger than the number of observational constraints used in previous studies. We used an a posteriori source term to validate various observation datasets including a set of measurements of atmospheric ¹³⁷Cs concentration, and to revisit the analysis of episodic deposition patterns over Japan.