Fluid flow in carbonate reservoirs is largely controlled by multiscale fracture networks. Significant variations of fracture network porosity and permeability are caused by the 3D heterogeneity of the fracture network characteristics, such as intensity, orientation and size. Characterizing fracture network heterogeneity is therefore essential in order to understand and predict fluid flow in fractured reservoirs, but this cannot be accomplished using only 1D data from wells, which is usually the only type of data available from the subsurface. To extend the 1D data to 3D data we commonly analyze the regional deformation, as different types of fold mechanisms produce different deformation styles and subsequent fracture patterns. 2D outcrop studies of fractures are often used to quantify these multi-scale relations between fracturing and large-scale structures. We build a geometric model, then make a mechanical analysis, followed by populating the fracture domains with outcrop-derived information. We use a novel approach called Digifract, which allows us to collect large amounts of 2D fracture data from outcrops, including fracture size, orientation and spacing measurements. Using this method we accurately quantify the links between multi-scale deformations, from fractures to regional tectonics. We applied the Digifract method in the outermost foothills of the Southern Tunisian Atlas, analysing fractures in different domains of four external folds with simple geometries and deformation histories. The lithology of the outcropping cores of all anticlines consists of the same fractured carbonates. The dimensions of the folds are on the same scale as reservoir analogs and form analogs for reservoirs in the Ghadames/Illizi basin, covering parts of Algeria, Tunisia and Libya. We relate differences in fracture characteristics to different localities (e.g., far or close to fold axes) and different folding stages in order to derive general rules that can be applied to subsurface fold analogs.