Bivariate drought recurrence analysis using tree ring reconstructions

Droughts may be represented by two main characteristics-duration and severity. In this paper, a general methodology to evaluate the frequency and risk of the occurrence of droughts is presented using a bivariate drought characterization. The theory of runs is applied to model drought recurrence as an alternating renewal process, describing droughts simultaneously in terms of their durations and severities. Short historical records usually do not allow reliable bivariate analyses. However, tree ring reconstructions of droughts provide information about past events, allowing the analysis. An approach to adapt and include dendrochronology reconstructions combined with historical records to characterize droughts is presented. The proposed approach uses the stochastic structure of the residuals of paleo reconstructions to generate equally likely representations of past drought events. The procedure was applied to paleo and historical records in Texas Climatic Division 5 and compared with univariate analyses. The application shows the bivariate analysis advantages in drought characterization.