Against the backdrop of global climate change, drought-flood abrupt alternation (DFAA) events have become increasingly frequent, yet the research on their driving mechanisms remains in the exploratory stage. To address the limitation of existing studies that focus primarily on the linear effects of climate change and atmospheric circulation, this study incorporated multiple factors, including surface energy fluxes, to conduct a multidimensional analysis. Using a revised DFAA index (R-SDFAI), we systematically analyzed the linear time-lag effects and nonlinear interactions of these factors on global DFAA across different lag times, employing Pearson correlation coefficients, multiple linear regression, and interpretable machine learning models. The study found that DFAA events were most frequent and intense in continental climate zones, whereas overall risk was relatively low in tropical climate zones. After accounting for time-lag effects, the explanatory power of multiple factors on DFAA increased from 33.03% to 70.05%, revealing clear spatial heterogeneity. For instance, in tropical climate zones, DFAA was bidirectionally influenced by vapor pressure deficit, whereas in arid climate zones, net radiation exhibited bidirectional associations with DFAA. After removing intra-annual seasonal signals, the dominant relationship converged on the moisture component. The study further revealed the key nonlinear threshold regulation of multiple factors, including the negative impacts of high net radiation and low rainfall in tropical climate zones, and the synergistic reversal driven by low heat flux and solar-induced chlorophyll fluorescence in arid climate zones. These findings transcended conventional linear frameworks for DFAA analysis, laying a scientific foundation for accurate prediction and disaster prevention.