Abstract Smoothing algorithms provide a means of identifying significant patterns in sets of orientated data, eliminating local perturvations within the observations and predicting patterns of orientated data in places which lack observations. Here we present the smoothing of orientation data with a distance-related method of data weighting as an alternative to previous weighting algorithms. The data weighting and smoothing method presented here in theory and practice is developed on the basis of a statistical smoothing algorithm. The method can be applied to orientation data of 180° periodicity such as maximum horizontal tectonic stresses (SH) as compiled in the World Stress Map database. Our smoothing algorithm enables discrimination between local (<250 km of lateral extent) and regional (c. 250–5000 km of lateral extent) stress fields, and allows comparison of SH with other directional data such as fault trends or strain data. We present smoothed stress maps for northeastern America, the Himalayas and western Europe. By varying the scale and smoothing parameters we illustrate their influence on the accuracy and smoothness. We give recommendations for the appropriate choice of these parameters.

Slab break-off is a plate-tectonic process which does not only return lithospheric material into the deeper mantle, but also has severe effects on surface movements and seismic hazard: slab-pull induced! seismicity is reduced when the subducted slab decouples from the overlying; crust. In the Vrancea region (SE Carpathians), strong earthquakes frequently occur at intermediate depths (70-180 km) in a laterally small volume, while the crust shows low but distributed seismicity. The stress pattern shows similar partitioning with vertical extension in the slab and no preferred, orientation in, the overlying crust. Both features indicate a decoupling between slab and overlying crust, either by slab break-off or delamination. However, the strong vertical elongation of the slab requires that the upper end of the slab is fixed vertically. Thus, a 'soft' coupling is assumed that is strong enough to enable elongation of the slab, but weak enough to inhibit 'quasi-static' stress transfer to the overlying crust.