Resilient water, food, and energy management strategies for an ever-growing population and changing environment depends on our understanding of water and carbon cycles from local to global scales. Fluxes of water and carbon are coupled by photosynthesis and plant transpiration cycles the largest fraction of terrestrial water from the land back to the atmosphere. Our limited ability to characterize interactions between hydrology and climate, regulated by plants’ response to stress, contributes to the greatest source of uncertainty in climate and carbon projections. Parameterized models need to represent the complexity and diversity of plant water use strategies, but hydrologically relevant model inputs are difficult to measure at ecosystem scales. Soil moisture integrates landscape fluxes and the spatial and temporal variability in soil moisture reflects dynamics of dominant land-surface processes. Diagnosing variability in soil moisture observations from point to landscape scales can thus quantify characteristics which are not measured directly. The central hypothesis of this dissertation is: soil moisture observations encode valuable ecohydrological information, and this information can be extracted to quantify plant water use strategies. This dissertation develops: (1) an inverse modeling framework to estimate scale-specific ecohydrological thresholds from probability distributions of soil moisture observations; (2) a global dataset of thresholds of soil water uptake, which are consistent with satellite soil moisture; and (3) relations between evapotranspiration and soil moisture at a range of biomes, based on the energy spectrum and probability distribution of soil moisture and information theory metrics. This work provides data driven methods that leverage new global observations and quantify ecohydrological relations which are critical to a variety of open climate, water, and ecosystem research questions and modeling endeavors.