Drought is a complex natural disaster phenomenon. It is of great significance to analyze the occurrence and development of drought events for drought prevention. In this study, two drought characteristic variables (the drought duration and severity) were extracted by using the Theory of Runs based on four drought indexes (i.e., the percentage of precipitation anomaly, the standardized precipitation index, the standardized precipitation evapotranspiration index and the improved comprehensive meteorological drought index). The joint distribution model of drought characteristic variables was built based on four types of Archimedean copulas. The joint cumulative probability and the joint return period of drought events were analyzed and the relationship between the drought characteristics and the actual crop drought reduction area was also studied. The results showed that: (1) The area of the slight drought and the extreme drought were both the zonal increasing distribution from northeast to southwest in Yunnan Province from 1960 to 2015. The area of the high frequency middle drought was mainly distributed in Huize and Zhanyi in Northeast Yunnan, Kunming in Central Yunnan and some areas of Southwest Yunnan, whereas the severe drought was mainly occurred in Deqin, Gongshan and Zhongdian in Northwest Yunnan; (2) The drought duration and severity were fitted the Weibull and Gamma distribution, respectively and the Frank copula function was the optimal joint distribution function. The Drought events were mostly short duration and high severity, long duration and low severity and short duration and low severity. The joint cumulative probability and joint return period were increased with the increase of drought duration and severity; (3) The error range between the theoretical return period and the actual was 0.1–0.4 a. The year of the agricultural disaster can be accurately reflected by the combined return period in Yunnan Province. The research can provide guidelines for the agricultural management in the drought area.

The matching degree between agricultural water and land resources directly determines the sustainable development of regional agriculture. Based on climate data corrected by delta statistical downscaling from five global climate models (GCMs) in the Coupled Model Intercomparison Project Phase 6 (CMIP6) and a multi-model ensemble, this study simulated the runoff used by the Variable Infiltration Capacity (VIC-3L) model under four emission scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) and analyzed the land use changing trend to obtain the matching degree between agricultural water and land resources. The results demonstrate that annual climate factors exhibit an increasing trend, and the average annual runoff was 2128.08–2247.73 × 10⁸ m³, during 2015–2100 under the four scenarios. The area of farmland changed with an increased area of 4201 km² from 1980 to 2020. The agricultural water and land resources would be well matched under the SSP1-2.6 and SSP2-4.5 scenarios in 2021–2100. However, the risks of mismatch would occur in the 2030–2040 and 2050–2060 periods under the SSP3-7.0 scenario, and the 2030–2040 and 2080–2090 periods under the SSP5-8.5 scenario. This study can provide insight into the scientific decision support for government departments to address the challenges of mismatching risks of agricultural water and land resources.