Ensemble-based data assimilation methods have shown amazing skill in reducing initial condition errors for numerical weather prediction. However, challenges still remain for basic ensemble Kalman filter algorithms to perform optimally when applied to highly nonlinear dynamical systems with complex relations between observations and states. Recently, many research groups have made progress in advancing methodologies for scenarios such as nonlinear and multi-scale dynamical systems, deviation from Gaussian distribution for state variables, and complex observation operators. The shear complexity of data assimilation and modeling systems often limits the communication and collaboration among researchers with different scientific foci. Yet, inter-disciplinary collaboration is an important step in cracking the nonlinear data assimilation problem. In this session, we welcome researchers to showcase their new findings in developing ensemble filtering methods that provide better results than basic ensemble Kalman filters for problems with large dimensions and strong nonlinearity. We also encourage authors that participate in this session to provide their insights on what should be the commonly adopted benchmark test cases that reflect the current data assimilation challenges. We envision that the discussion stemming from this session will contribute to wider collaboration among researchers to push forward the frontiers of data assimilation science.