Meteorological drought (MD) is the source of agricultural drought (AD), and in-depth understanding of the propagation process from MD to AD can help AD early warning. However, previous studies primarily focused on average state of propagation time (PT) rather than from the perspective of various drought levels, and had coarse spatiotemporal resolutions (monthly and basin scales), and also ignored the PT during critical stages of crop water demand. To fill the knowledge gap, this study used the standardized precipitation index (SPI) and standardized soil moisture index (SSMI) to characterize MD and AD. Then, a framework for high-resolution PT identification under different drought levels based on Copula theory and conditional probability was proposed. Taking the rain-fed agricultural region of Loess Plateau (RFLP) as an example, the PT of the critical stages of water demand for local crops (spring maize, winter wheat) was calculated respectively. Result showed that: (1) PT in summer and autumn is shorter than those in winter and spring, and the higher drought level, the faster the spreading rate, in which the PT in spring, summer, autumn, and winter under extreme MD conditions were 3– 15, 1– 6, 1– 8 and 4– 16 ten days, respectively; (2) it should be noted that the central part of RFLP is a hotspot, with a shorter PT than any surrounding area; (3) PT is mainly related to local water and heat characteristics, and high potential evapotranspiration tends to accelerate the propagation process, while sufficient soil moisture has a good buffering effect on MD and can slow down the propagation; (4) it was observed that PT in the jointing-heading stage of wheat was significantly shortened, which should be of great concern to the local agricultural authorities. In general, this study sheds new insights into high spatiotemporal resolution drought propagation, which is helpful for AD risk management.