Water availability critically influences ecosystem productivity, particularly in arid environments. Observed responses of vegetation to both spatial gradients and temporal variability in water availability have been used to better understand vegetation-water relationships. Our study investigates the divergent vegetation responses to water availability across space and time, critical in forecasting responses to climate change. Using soil moisture data from NASA’s SMAP, we employ boundary line analysis and quantile regression to understand the complex relationship between soil water availability and Gross Primary Productivity (GPP). This enables us to isolate GPP's direct response to soil water changes, mitigating other factors like temperature and vapor pressure deficit.

Our results for the Contiguous United States reveal that vegetation exhibits greater spatial sensitivity to water availability compared to temporal sensitivity. This distinction becomes less pronounced when assessing water availability through soil moisture rather than precipitation. We identify soil field capacity and average soil moisture levels as crucial drivers of these sensitivities, demonstrating the soil's important buffering role.

Our findings suggest that land surface models greatly underestimate the GPP response to soil moisture in arid regions while overestimating it in humid regions. They also inadequately capture spatial GPP variability along soil moisture gradients, indicating limited soil-plant-atmosphere representation in current models, and raising significant concerns for future carbon projections.

Our research underscores the critical soil buffering role in mediating ecosystem responses to water availability changes, further highlighting the need for improved models. This study aids in understanding complex ecohydrological processes, providing soil moisture-based benchmarks for model accuracy and projecting climate change impacts on arid ecosystems.