Seismicity on large faults is often characterized in terms of an independent recurrence time and magnitude distribution, which forms the basis for calculating future earthquake probabilities. The underlying assumption is that the driving mechanism for earthquakes on any particular fault is uniquely linked to that fault, determined by the rate of long-term creep on its deeper extension. However, our modelling of nearly 20 years of Global Positioning System data along the obliquely converging plate boundary in New Zealand shows that interseismic stress accumulation can be independent of the properties of the numerous crustal faults and controlled by locking on the megathrust. In this way, the interseismic driving mechanism for large crustal earthquakes is not linked directly to the individual major faults that rupture. This scenario predicts large-magnitude earthquakes with complex multifault ruptures in broad zones that ‘jump’ from fault to fault, following the contours of stress/strain loading. This can explain the November 2016 M_w7.8 Kaikoura earthquake that shattered the plate boundary in central New Zealand. Repeated episodes of this would create the observed complex array of active faults with the appearance of coherent slip. Our analysis opens up the possibility to use long-term Global Positioning System data to identify this type of earthquake behaviour.

大断层的地震活动通常以独立的复发时间和震级分布为特征，这构成了计算未来地震概率的基础。基本假设是，任何特定断层的地震驱动机制都与该断层有独特的联系，这取决于其深层扩展的长期蠕变速率。但是，我们对新西兰倾斜汇聚板块边界近20年的全球定位系统数据进行的建模显示，地震应力累积可以独立于众多地壳断层的性质，并可以通过锁定大推力来控制。这样，大地壳地震的震源驱动机制就不会直接与个别破裂的主要断层联系在一起。该场景预测了在应力/应变载荷等高线的情况下，在断层之间“跳跃”的宽阔区域发生复杂的多断层破裂的大地震。这可以解释2016年11月的M_w 7.8凯库拉地震破坏了新西兰中部的板块边界。重复这样的过程将形成观察到的活动断层的复杂阵列，并出现连贯的滑动。我们的分析为使用长期全球定位系统数据识别这种地震行为开辟了可能性。