Forests account for over 90% of total biomass globally and are absorbing over a quarter of anthropogenic CO2 emissions every year. Forest dynamics, such as growth, mortality, and recruitment, is critical to predicting the future fates of terrestrial biosphere under a changing climate. Predictive modeling of forests as a dynamical system has made various progresses in the past decades. However, much of the uncertainty in model structure and parameters arise from the lack of process-informative observations, which are usually labor- and time-intensive. In this seminar, I will present some past and on-going work to (1) monitor tropical forest mortality at landscape scale by integrating limited ground observations and big data from satellite remote sensing (2) monitor individual-level tree growth using dense 3D point cloud from terrestrial lidar. The mortality project borrowed ideas from the critical transition theory while the lidar project involved computational methods to deal with massive 3D position data.