Stepwise clustering future meteorological drought projection and multi-level factorial analysis under climate change: A case study of the Pearl River Basin, China

Climate change has significant impacts on the Pearl River Basin, and the regional ecological environment and human production may face severe challenges in the future due to changes in temperature and precipitation, as well as their derivative disasters (e.g., drought). Therefore, a full understanding of the possible impacts of climate change on Pearl River Basin is desired. In this study, the potential changes in temperature, precipitation, and drought conditions were projected through a stepwise clustering projection (SCP) model driven by multiple GCMs under two different RCPs. The developed model could facilitate specifying the inherently complex relationship between predictors and predictands, and its performance was proven to be great by comparing the observations and model simulations. A multi-level factorial analysis was employed to explore the major contributing factors to the variations in projecting drought conditions. The results suggested that the Pearl River Basin would suffer significant increasing trends in Tmean (i.e., 0.25?0.34 ?C per decade under RCP4.5 and 0.42?0.60 ?C per decade under RCP8.5), and the annual mean precipitation would increase under both RCPs. The drought events lasting for 1?2 months would be decreased by 7.7%, lasting for 3?4 months would be increased by 4.3%, and lasting for more than five months would be increased by 3.4% under RCP4.5, respectively. While they changed to 6.1%, 1.4%, and 4.7% under RCP8.5, respectively. More medium and long-term drought events with higher drought severity would occur. GCM has dominant influences on four different responses of drought duration, accounting for 50.20%, 52.61%, 56.71%, and 56.24% of total variabilities, respectively. Meanwhile, the effects explained by GCM*RCP interactions cannot be neglected, with an average contribution rate of 44.37%, 37.86%, 37.66%, and 35.83%, respectively.