Future Climates from Bias-Bootstrapped Weather Analogs: An Application to the Yangtze River Basin

Abstract

Abstract The authors describe a statistical analog resampling scheme, similar to the “intentionally biased bootstrap,” for future climate projections whose only constraint is a prescribed linear temperature trend. It provides a large ensemble of day-to-day time series of single-station weather variables and other climatological observations at low computational cost. Time series are generated by mapping time sequences from the observed past into the future. The Yangtze River basin, comprising all climatological subregions of central China, is used as a test bed. Based on daily station data (1961–2000), the bootstrap scheme is assessed in a cross-validation experiment that confirms its applicability. Results obtained for the projected future climates (2001–40) include climatological profiles along the Yangtze, annual cycles, and other weather-related phenomena (e.g., floods, droughts, monsoons, typhoons): (i) the annual mean temperature and, associated with that, precipitation increase; (ii) the annual cycle shows an extension of the Asian summer monsoon season with increasing rainfall, linked to a small summer temperature reduction in the Yangtze lower reaches; (iii) coupling between monsoon circulation and monsoon rainfall strengthens; (iv) while drought occurrence is reduced, Yangtze floods do not change considerably; and (v) the number of typhoon days in the East China Sea shows a reduction of about 25%; the proportion of intense typhoons with landfall increases. GCM scenario simulations produce similar results.