Tucker 2025, MULTI-SCALE DRIVERS OF INTRA-CATCHMENT VARIABILITY IN DROUGHT RESPONSE: LINKING REMOTE-SENSING, GEOPHYSICAL, AND ECOHYDROLOGICAL DATA

Quantifying fine-scale controls on tree-stand drought response remains challenging due to the interacting effects of landscape, moisture availability, and vegetation. We investigated drought resistance—the ability of a forest to continue transpiring during drought—and resilience—the ability to rebound post-drought—in a 0.5 km2 subcatchment of the Southern Sierra Critical Zone Observatory (King’s River Experimental Watershed), during and after the 2012-2016 California drought. Using a multi-scale approach, we integrated catchment-wide remote sensing data (LiDAR and Normalized Difference Vegetation Index, NDVI) with tree-scale in situ ecohydrological (sapflow and soil moisture), meteorological (air temperature and vapor pressure deficit), and geophysical (electrical resistivity) data from six stations. We fitted generalized additive models (GAMs), which capture nonlinear relationships, using eight remote-sensing-derived predictors—elevation, slope, aspect, distance to stream, topographic wetness index (TWI), snow depth, canopy height, and baseline NDVI. The intercorrelated variables of elevation, slope, distance to stream, TWI, and baseline NDVI were the strongest predictors of resistance and resilience. Notably, baseline NDVI had approximately the opposite effects on resistance and resilience, highlighting the need to distinguish between drought impacts during versus after drought events. The GAMs explained 51% of the variance in resistance and 39% in resilience, indicating that additional covariates are needed, potentially at the finer, plant-scale. Our in-situ data from the valley bottom indicated the presence of hydrologic refugia—areas that retain higher soil moisture than surrounding terrain—which helped explain the added drought resistance observed there. By combining our ecohydrology and geophysical data, we tracked when trees sourced water from sources other than their root zone (such as internal tree water stores) and identified changes to the active rooting zone extent on the sub-daily scale. During seasonal water deficits, trees accessed deeper rather than broader water stores and increasingly relied on internal reserves until depletion. Stations where internal stores were exhausted more rapidly exhibited lower drought resistance. Altogether, this work emphasizes that accurate predictions of forest drought response—especially in complex montane watersheds—require attention to multi-scale processes.

由于景观、水分有效性与植被间的交互作用，量化林分干旱响应的精细尺度调控机制仍具挑战性。本研究针对南内华达临界带观测站（Southern Sierra Critical Zone Observatory，国王河实验流域）一处0.5平方千米的子集水区，在2012-2016年加州干旱期间及灾后，探究了干旱抗性——森林在干旱期间维持蒸腾作用的能力——与恢复力——干旱后恢复至基准状态的能力。本研究采用多尺度研究方法，整合了流域尺度的遥感数据（激光雷达（LiDAR）与归一化植被指数（Normalized Difference Vegetation Index, NDVI）），以及6个监测站点的树木尺度原位生态水文数据（树液流量与土壤湿度）、气象数据（气温与水汽压亏缺）与地球物理数据（电阻率）。本研究拟合了可捕捉非线性关系的广义加性模型（generalized additive models, GAMs），选用8项遥感衍生预测变量：海拔、坡度、坡向、距溪流距离、地形湿度指数（topographic wetness index, TWI）、积雪深度、冠层高度与基准NDVI。其中海拔、坡度、距溪流距离、TWI与基准NDVI这些互相关联的变量，是干旱抗性与恢复力的最强预测因子。值得注意的是，基准NDVI对干旱抗性与恢复力的影响大致相反，这凸显了区分干旱发生期间与灾后干旱影响的必要性。GAMs分别解释了51%的干旱抗性变异与39%的恢复力变异，表明仍需纳入更精细的植物尺度的额外协变量。我们在谷底获取的原位数据显示，存在水文避难所——即土壤湿度高于周边区域的地带——这有助于解释当地观测到的更高干旱抗性。通过结合生态水文与地球物理数据，我们追踪了树木何时从根区以外的水源（如树木内部储水）获取水分，并识别了亚日尺度下活性根区范围的变化。在季节性水分亏缺期间，树木会转向深层而非更广范围的水源，并愈发依赖内部储水直至耗尽。内部储水耗尽更快的监测站点，其干旱抗性更低。综上，本研究强调，若要准确预测森林的干旱响应——尤其是在复杂的山地流域中——需关注多尺度过程。