Using daily temperature to predict phenology trends in spring flowers

摘要

The spring season in Korea features a dynamic landscape with a variety of flowers blooming sequentially one after another. This enables local governments to earn substantial sightseeing revenues by hosting festivals featuring spring flowers. Furthermore, beekeepers move from the southern tip of the Korean Peninsula all the way northward in a quest to secure spring flowers as nectar sources for a sustained period of time. However, areal differences in flowering dates of flower species are narrowing, which has economic consequences. Analysis of data on flowering dates of forsythia (Forsythia koreana) and cherry blossom (Prunus serrulata), two typical spring flower species, as observed for the past 60 years at six weather stations of the Korea Meteorological Administration (KMA) indicated that the difference between the flowering date of forsythia, the earliest blooming flower in spring, and cherry blossom, which flowers later than forsythia, was 14 days on average in the climatological normal year for the period 1951-1980, compared with 11 days for the period 1981-2010. In 2014, the gap narrowed further to 7 days, making it possible in some locations to see forsythias and cherry blossoms blooming at the same time. Synchronized flowering of these two flower species is due to acceleration of flowering due to an abnormally high spring temperature, and this was more pronounced in the later-blooming cherry blossom than forsythia. While cherry blossom flowering dates across the nation ranged from March 31 to April 19 (an areal difference of 20 days) for the 1951-1980 normal year, the difference ranged from March 29 to April 12 (an areal difference of 16 days) for the 1981-2010 normal year, and in 2014, the flowering dates spanned March 25 and March 30 (an areal difference of 6 days). In the case of forsythia, the gap was narrower than in cherry blossoms. Climate change in the Korean Peninsula, reflected by rapid temperature hikes in late spring in contrast to a slow temperature rise in early spring immediately after dormancy release, likely brought forward the flowering date of cherry blossom. We derived a thermal time-based flowering model from this analysis and used it to predict the flowering dates of forsythia and cherry blossom in 2014. The root mean square error for the prediction was within 2 days from the observed flowering dates in both species, showing a feasibility of prediction under the changing climate.