Prediction of Ambient Dose Rate Based on Linearization and Cross-validation Method

Abstract

We propose a linearization and cross-validation method to make long-term prediction of the ambient dose rate in the environment. The time-series of gamma-ray flux is calculated by using the decay of two radionuclides (¹³⁴Cs and ¹³⁷Cs) and the carrier diffusion. The linearization and cross-validation method are adopted for forecasting steps. Under the linearization, a prediction is reduced to a linear extrapolation. Using the cross-validation method, we can compare the calculated series with the observations, and check the precision. An ability-index to execute these processing is shown.

We test the approach on use of observation data in the monitoring-post at Shimo-Oguni meeting center in Date-city of Fukushima. It gives an observation tracing curve over 0.9 as the determination constant, and predicts an ambient dose rate about 0.7 during next 1 year period.

We assume people who were evacuated toward other municipalities due to the Fukushima Nuclear Power Plant Accident, and who conduct the timing to return back home now.