The goal of our research is to understand the complex relationship between the duration and size of forest fires so as to better predict these important characteristics of fires not only for the purpose of fire management but also for a fundamental understanding of fire science. In our case, we jointly model time spent (duration) in days and area burned (size) in hectares from ground attack to final control of a fire as a bivariate survival outcome using a joint modeling framework that connects the two outcomes with a shared random effect. The talk will focus on discussing: (1) the robustness of joint models, where the true model is a copula, (2) how duration and size are related in terms of their shared error, (3) the implementation of the optimal framework developed as a component in the fire prediction system that is concurrently under development by Natural Resources Canada.