Abstract
The long-term series of the omiwatari record compiled by Fujiwara and Arakawa (1954) is a prominent database for climate reconstruction, especially for winter severity in Japan. The omiwatari phenomenon consists of cracks on the frozen surface of Lake Suwa (36°02' N, 138°05' E). It is deemed to a signal for predicting the harvest in the coming summer, and therefore the date of freezing has been recorded by various agencies.
The time series begun in 1444 consists of records in several different documents and continued to be recorded by the local Suwa Meteorological Observatory until a few years ago. The contents of the documents differ and there may be some heterogeneity in the database. It is confirmed in this study that the definition of omiwatari could be changed. Although the occurrence of the phenomenon is determined only through daytime observations today, the time of occurrence at night were also recorded in the historical records. Moreover, the average minimum air temperature on and one day before the freezing date differs statistically between freezing dates observed by the local Suwa Meteorological Observatory and by the Suwa-Taisha Shrine. On the other hand, a similar statistical test indicates consistency in the freezing dates observed by the Suwa-Taisha Shrine during the 20th century.
The database has been already analyzed by several researchers. Taking into account the heterogeneity, however, climate reconstruction using the omiwatari database would be more reliable.
