Improved model for the predictions of Japanese cedar pollen dispersion and counts in Fukushima City

Abstract

Japanese cedar pollinosis is the most common seasonal disease in Japan and has been considered a national affliction. The predictions of the dispersion and counts of Japanese cedar pollen are important for self- and medical care. We previously reported models for predicting the first days of pollen dispersion and total pollen counts during the pollen dispersal seasons based on meteorological data for the past 20 years. However, there could have been changes in factors related to the environment, climate, and forest, among others. Therefore, we reconsidered that the use of data from the recent 10 years, rather than the past 20 years, would be more suitable for accurate predictions and verified the results obtained by the previous and newly constructed models. Regarding the predictions of the first days of pollen dispersion from 2011 to 2020, the mean differences between the actual day and the days predicted by the previous and new models were 2.40 and 2.00 days, respectively. Retrospective investigation of the total pollen counts from 2012 to 2022 showed that the correlation coefficients for the actual counts and those predicted by the new model (r=0.962) were higher than those for the previous model (r=0.922). In summary, the new prediction models based on the data collected over the recent 10 years more accurately predicted the first day of pollen dispersion and total pollen count than the previous models based on the past 20-year data. Therefore, prediction models should be updated periodically.