This document describes the Po-tree, a new indexing structure for spatiotemporal databases with soft real time constraints. This is crucial for risk management, particularly for the monitoring of volcanic risk. For example, the Popocatepetl, a Mexican volcano, is closely monitored through an network of spatially referenced sensors. They all send back their measurements to a central spatiotemporal database. The Po-tree shall therefore be able to index huge quantities of data in a restricted amount of time. It shall also be able to favor the newer data, which are queried more frequently. While most structures tend to favor the temporal aspect over the spatial, or tend to consider time as another spatial dimension, the Po-tree offers an alternative. It does so by combining two different structures, for spatial and temporal dimensions respectively, stressing the spatial aspect, and linking every source of information to a new temporal sub-tree. While mobility is not yet supported, experimentations and comparisons have shown interesting potential in the structure.