Abstract:[Objective]In order to quantitatively analyze the response mechanism of agricultural drought to single factor in karst areas,[Methods]In this study, Delaunay triangulation and K-means clustering were employed to calculate the area and classify the annual rainfall data of 83 stations in Guizhou Province, resulting in the division of Guizhou into 5 regions. Subsequently, a vector autoregressive (VAR) model was established based on the meteorological factors including precipitation (Pre), average temperature (T), potential evapotranspiration (E), wind speed (W), air pressure (P), soil moisture content (SW), and Standardized soil moisture index (SSI) for each region. The response mechanism of agricultural drought to various time scales of meteorological factors and soil moisture content was analyzed using pulse response and variance decomposition.[Results] (1) The agricultural drought was higher in the west and lower in the east, and gradually decreased from south to north;The agricultural drought is more severe in spring and summer than in autumn and winter. . (2) The time required for the impulse response of each variable on the annual scale to reach the peak of SSI is Pre < E = W < P < SW < T, and the recovery period is as follows: Pre > W > T > E> P> SW. From the monthly scale to the annual scale, in areas with strong karst development, the time required for SSI pulse to reach its peak value is shortened by 2 periods for factor E; while SW is opposite to this, and the non-karst area has a shorter recovery period than karst area.(3) Rainfall plays a dominant role in agricultural drought, and rainfall and wind speed played a stronger role in agricultural drought with the increase of time scale. The contribution ratio of agricultural drought was 31.88%, 34.25%, 58.33%, 1.07%, 3.15% and 12.88%, respectively [Conclusion]The research results can provide a theoretical basis for the early warning of agricultural drought in karst areas.