Drought has been one of the most serious and long-term natural disasters, leading to a significant impact on production and life. A separation of stripping time or space can be usually adopted to simplify the drought assessment at present. But the spatiotemporal continuous characteristics of drought cannot be fully considered during the assessment, because the separate analysis is limited to time and space. At the same time, there is an upward trend of drought in the humid region of South China in recent years. Therefore, it is highly urgent to explore the evolution of drought under the background of climate change. This study aims to determine the spatiotemporal variation of future droughts in the Pearl River flow region of South China. A coupled model comparison project phase 6 (CMIP6) data was firstly selected in the BCC-CSM2-MR precipitation dataset under the rliplfl experimental mode. The standardized precipitation index (SPI) was then calculated to analyze the overall drought trend under the scenarios of SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5. Secondly, the Sen's slope was used to qualitatively analyze the future drought trend in the study area. The Hurst index was also used to comprehensively describe the persistence of the future drought trend. Thirdly, a rotating empirical orthogonal function (REOF) was utilized to spatially classify the study area, according to the dry and wet characteristics. The spatial characteristics of the future drought index were then obtained to compare the different spatial modes under different climate scenarios. Finally, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) was combined to decompose the drought duration and frequency of different SPI spatial modal divisions. As such, the spatial-temporal evolution characteristics were achieved in the future drought. The results showed that: 1) The BCC-CSM2-MR precipitation dataset was suitable for the SPI calculation under the CMIP6 and observation data. 2) Five SPI divisions were divided to decompose by REOF. The divisions depended mainly on the radiative forcing, but there was an almost unchanged area of each division. 3) There was a much more humid state of divisions in the coastal space, while the consistent changes of different dry and wet divisions, according to the future trend of SPI. 4) There were shorter periods of drought duration and frequency after CEEMDAN analysis, indicating the more significant characteristics in the short period of coastal, spatial and modal zoning. 5) There was a much more serious drought with the increase of radiative forcing under different climate models using the CEEMDAN analysis. Specifically, the severe drought was last for about 0.5-1.8 months in the future, while the drought frequency was between 5.6% and 13.9%. Consequently, this finding can provide a new perspective of spatiotemporal coupling analysis for drought research, particularly for drought monitoring, prediction and management.