The analysis of large movement datasets is a challenging task, because of their size and spatial complexity. This paper presents an interactive geovisual analytics approach named Hybrid Spatio-Temporal Filtering that integrates filtering of multiple movement characteristics, geovisual representations of the data, and multiple coordinated views to enable analysts to focus on movement patterns that are of interest. In particular, we propose a novel technique that combines the fractal dimension and velocity of movement paths to filter out uninteresting records through an iterative signature-building process. In order to allow analysts to explore the data at different scales of the movement path length, fractal dimension estimation is performed using an adjustable moving window technique. These tools are provided in conjunction with a probability-based zonal incursion tool to visually represent when the movement nears areas of interest. The outcome is a geovisual analytics system that allows analysts to specify a hybrid filter consisting of the desired movement path complexity, the length of the paths to consider, and the velocity range that represents specific types of behaviors. This filtering of the data supports analysts in identifying movement paths that match their specified interests, resulting in a reduction in the amount of data shown to the analyst. The utility of the approach was validated through field trials, wherein fisheries enforcement officers analyzed and explored fishing vessel movement data using the prototype system. The participants responded positively to the features of the system and the support it provided for their data analysis activities. The combination of fractal dimension, velocity, and temporal filtering helped them to effectively identify subsets of data that conformed to particular behavioral patterns of interest.