Projections on climate internal variability and climatological mean at fine scales over South Korea

Abstract

Climate internal variability (CIV) plays an important role in understanding climate and is one of the principal uncertainties in climate projections. This study aims to estimate CIV and climatological mean (CM) in predictions using different emission scenarios for South Korea. A stochastic weather generator is employed to generate 100 ensembles of 30-year hourly time series for 40 meteorological stations. CIV is then estimated from the detrended method and compared with the noise computed by the two approaches. The extremely high value of the coefficient of determination between CIV values and noise indicates that the methodologies are seamless. The key results of this study include: (1) national average CM and CIV will increase in the future, and that increase will be greater in Representative Concentration Pathway 8.5 and end periods; (2) the nature of future changes in CM and CIV differ according to the indices of interest. Characteristics of three precipitation-quantity indices (total precipitation, totPr; daily maximum precipitation, maxDa; and hourly maximum precipitation, maxHr) and the precipitation-occurrence index (the number of days without precipitation, nonPr) are largely distinct; (3) examining the relationship between factors of changes of CIV and CM reveal a high correlation between them for maxDa and maxHr, but not for other indices; (4) The tail information of distribution for the FOC ratio implies that future changes in total and extreme precipitation are likely to be decoupled for some months or at some locations. The degree of decoupling is more noticeable on the hourly than the daily scale; and (5) the spatial deviation of CIV is also larger during the summer when CIV values are spatially large; this is valid only for totPr and maxDa. Methodologies and results for finer scales help assess the impact of climate change and develop appropriate adaptation and response strategies.