A comprehensive understanding on ecosystem cycles, which is crucial for decisions and policies in climate protection, can only be possible when the environments and key ecosystems are fully and continuously tracked. This project demonstrates a solution using high temporal and spatial resolution terrestrial laser scanning (TLS) time series. The main aim is to develop an edge computing based pipeline to automatize the collection and analysis of big point cloud time series data, which is constantly produced at a rate of ~10 GB/hour. This enables the modeling of the relationship between the dynamics of tree-level Gross Primary Production (GPP) and tree, stand, and local weather conditions. This also facilitates a long term self-learning process that can automatically and continuously optimize the models with growing data. Such model can support decision making in climate protection by providing prediction of GPP dynamics caused by changes in forest stands and weather conditions.