Journal of Computer Chemistry, Japan
Online ISSN : 1347-3824
Print ISSN : 1347-1767
ISSN-L : 1347-1767

This article has now been updated. Please use the final version.

Prediction of Ambient Dose Rate Based on Linearization and Cross-validation Method
Tomoo AOYAMAToru YAGIJunko KAMBE
Author information
JOURNAL FREE ACCESS FULL-TEXT HTML Advance online publication
Supplementary material

Article ID: 2016-0070

Details
Abstract

環境中のγ線線量値を長期にわたり予測するため,放射性核種の物性と担体の拡散を考慮した空間線量値の時間変化を直線化する方法を提示した.現象の変化が直線化できると予測は線形補外となる.交差検証法で過去のデータの一部を用いて予測を行い実測との対応から予測精度を見積る.将来の予測計算も,過去の処理と同一である.方法自体の精度のオーダは同じである.本処理の可能性を示す指標も示した.

福島県伊達市下小国集会所のモニタリング・ポストのデータでテストすると,線量予測式と実測値の決定係数は0.9以上,1年先の予測値では0.7であった.観測点の気象的擾乱,周辺環境の変化は決定係数の低下として現れる.

我々は,福島原子力発電所事故により他の市町村に避難し,帰宅時期を考慮している人を想定した.

Figures
Figure 1.

 Vector representation of β(t) function.

Vertical axis is the value of β(t) function in unit [1/d].

Horizontal axis is the elapsed day from March 15, 2011. Dots correspond to β(t) function per 15 [d], and straight line is the regression one.

Figure 2.

 Prediction and observation curves from 2012 to 2019.

The vertical axis is the ambient dose rate [μSv/h] and the horizontal is elapsed days from March 15, 2011. The blue zigzag are observation dose rates, and the red is fitting (until 2016) and prediction (after 2017).

Figure 3.

 Fitting and observations of low R2 period in Table 1.

Blue dots are observations. Red curve is calculated by Eq. (11).

Table
Table 1. Prediction and the precision based on determination coefficients
Observation PeriodEffective**/total daysFitting R2Prediction R2
1B*~2012.12.10255/2550.8720.759
2B~2013.12.10583/6200.9560.784
3B~2014.12.10875/9850.9820.413
4B~2015.12.101217/13500.9860.636
5B~2016.12.91559/17150.909Figure 1

*) B is the start day that is 2012.3.31.

**) Effective days is that no-measurement and snow days are subtracted from total days.

R2 is calculated for the effective days in Linear{dose (t)} set.

Prediction R2 is calculated among predicted dose rates and observations. The observation period is from 2012.3.30 to 2016.12.9.

文献
 
© 2017 Society of Computer Chemistry, Japan
feedback
Top