Daily precipitation and mean temperature data of 18 meteorological stations over the Beijiang River Basin of Guangdong for the last fifty years were analyzed to classify anomalously dry conditions using the standardized precipitation and evapotranspiration index (SPEI). In this study，drought variables (drought duration

drought severity and drought severity peak) were identified using run-length theory

11 probability distribution functions were adopted to analyze the probability behaviors of drought severity and drought severity peak at 18 meteorological stations. The L-moment technique is used to estimate the parameters of the probability functions

the Kolmogorov-Smirnov method (K-S) is accepted to evaluate the goodness-of-fit of the probability functions and evaluate the tenyear return periods values. Based on the Genest-Rivest method and the tail dependency of random variable

GH copula that has up tail dependency is used for drought variable connect function. The Kendall rank correlation coefficient method is used to estimate the parameters of GH copula function. Then we comprehensively analyze the spatial variability of joint probability distribution of two indices. The spatial variability of probability distribution of one drought variable and joint distribution of two drought variables can provide the scientific basis for analysis of the local water volume balance and reduce drought disaster risk.