Technological developments allow for gathering and processing an increasing amount of data in real time. The integration of these tools into risk assessment allows for the development of dynamic risk assessment and datadriven decision support. The latter is of special interest for systems that are remotely monitored and controlled by operator, such as power grids. Generally, grid operators have little access to environmental information to support decisions on interventions and preventive maintenance. Recent initiatives aim at integrating machine learning and other techniques into dynamic risk assessment of power grids. The performances of these initiatives depend on the quantity and quality of data one can gather and process, the available technology, and the cost-benefit ratio of which these initiatives are synonym. In addition, the development of these solutions must be completed by the list of decisions to which the operators may be subject, as well as the information required in order to make the correct decisions for the system's needs. This paper presents a framework for optimizing decision-support of power grids operators using data-driven solutions, focusing on risks associated to vegetation. We analyse the possible scenarios concerning power grids under risk by surrounding vegetation, and the deriving decisions the operators can make under those scenarios. We further analyse and discuss the information required by the operators for decision making. This information is finally integrated into a data-collection and processing framework.