Changes in soil moisture strongly affect vegetation growth, which may in turn feed back on soil moisture by directly affecting evapotranspiration and indirectly regulating precipitation. Previous studies often focused on the unidirectional effects of soil moisture on temporal vegetation dynamics, yet bidirectional dependencies have rarely been studied. Here we analyzed the bidirectional dependency between soil moisture and vegetation productivity during 2001–2020 at a global scale using the Granger causality, and revealed strong concurrent and lagged correlations between both variables in large areas globally. Bidirectional causality between soil moisture and vegetation productivity was identified over 66% of the vegetated land areas, while considering lagged effects and controlling for temperature and solar radiation. Unidirectional effects of vegetation productivity on soil moisture, and soil moisture on vegetation productivity, were observed for 22% and 12% of vegetated land areas, respectively. For areas characterized by uni- and bidirectional dependencies, 74% of the vegetation productivity and 48% of soil moisture could be explained by optimum lag models. Finally, we observed increases in both vegetation productivity and soil moisture in 44% of the vegetated land areas, yet 36% showed an increase in vegetation productivity but a decrease in soil moisture, indicating divergent responses between greening and water availability. Identification of areas showing Granger causality between soil moisture and vegetation productivity is important for our understanding of carbon-water interactions for terrestrial ecosystems under climate change and for improving sustainable management of ecosystem services linked to the carbon-water cycle.

土壤水分的变化强烈影响植被生长，植被生长又可能通过直接影响蒸发蒸腾和间接调节降水来反馈土壤水分。以前的研究通常侧重于土壤水分对时间植被动态的单向影响，但很少研究双向依赖性。在这里，我们使用 Granger 因果关系分析了 2001-2020 年全球范围内土壤水分和植被生产力之间的双向依赖性，并揭示了全球大面积两个变量之间的强烈并发和滞后相关性。超过 66% 的植被土地面积确定了土壤水分和植被生产力之间的双向因果关系，同时考虑了滞后效应并控制了温度和太阳辐射。分别在 22% 和 12% 的植被土地面积上观察到植被生产力对土壤水分和土壤水分对植被生产力的单向影响。对于具有单向和双向依赖性的区域，74% 的植被生产力和 48% 的土壤水分可以通过最佳滞后模型来解释。最后，我们观察到 44% 的植被土地面积的植被生产力和土壤水分均增加，而 36% 的植被生产力增加但土壤水分减少，表明绿化和可用水量之间存在不同的反应。