Potential of Crowdsourcing Approach on Monitoring Radioactivity in Fukushima Prefecture

摘要

Crowdsourced data and professional scientists’ data were compared for agreement regarding air dose rate levels and trends in air dose rate reduction to assess the value of public initiatives in radiation data collection during nuclear crisis response. This study used seven KURAMA datasets from seven survey periods to represent expert group data. To represent non-expert group data, we used seven datasets from SAFECAST’s database the collection period of which was comparable to the KURAMA survey periods. A simple linear regression model was separately applied to a pair of combination datasets from different sources and also to a pair of first survey period datasets and subsequent datasets from the same source. The R-squared of the models showed the non-scientist group data correlating well with the corresponding expert data. The slopes of all the regression models, however, indicated that the air dose rate values measured by non-expert group were about 40 to 70 percent lower than those of the expert group. The air dose rate reduction trend from the crowd data showed a similar decreasing pattern compared to that of the expert group, although the discrepancy in the magnitude of dose reduction between them was as high as 14 percent. The discrepancy in air dose rate values suggest a careful interpretation of radiation information generated solely from crowdsourced data. Nevertheless, given the strong linear relationship of crowd data with scientist groups’ data, the superior number of data points during a crisis, and the flexibility and agility of self- organization, we argue that the public could be a great partner to scientist groups in radiation data collection.