|dc.description.abstract||Spatiotemporally continuous estimates of the hydrologic cycle are often generated through hydrologic modeling, reanalysis, or remote sensing methods, and commonly applied as a supplement to, or a substitute for, in-situ measurements when observational data are sparse or unavailable. Increased access over the past decade to distributed cloud systems and high-performance clusters have greatly increased the computational processing power available to the hydrologic community. In turn, the number of these modeled products estimating various components of the hydrologic cycle over continental and global extents have grown year by year. Despite this, significant gaps in our knowledge of terrestrial hydrology remain. Models typically host high levels of uncertainty in many parts of the conterminous United States (CONUS). Classic model validation studies are insufficient in describing model skill over such large extents due to limitations in data availability. Further, models are increasingly more expensive to operate and analyze due to higher levels of complexity in terms of internal equations, parameter requirements, differences in meteorological forcings, and calibration methodologies. Hydrologic and reanalysis models, capable of retrospective and forecast prediction, are spatially constrained to moderate or coarse resolutions. Remote sensing models, capable of fine resolution estimates of individual hydrologic components, are temporally constrained by availability of satellite observations. This dissertation first seeks to better quantify the uncertainty in modeled estimates of precipitation (P), actual evapotranspiration (ET), runoff (R), snow water equivalent (SWE), and soil moisture (SM) from 87 unique datasets generated by 47 hydrologic models, reanalysis datasets, and remote sensing products across the conterminous United States (CONUS). Uncertainty between hydrologic component estimates was shown to be high in the western CONUS, with median uncertainty (measured as the coefficient of variation) ranging from 11-21% for P, 14-26% for ET, 28-82% for R, 76-84% for SWE, and 36-96% for SM. Uncertainty between estimates was lower in the eastern CONUS, with medians ranging from 5-14% for P, 13-22% for ET, 28-82% for R, 53-63% for SWE, and 42-83% for SM. Study results show that disagreement between estimates can be substantial, sometimes exceeding the magnitude of the measurements themselves. The authors conclude that multi-model ensembles are not only useful, but are in fact a necessity, to accurately represent uncertainty in research results. Following this inter-model analysis, uncertainty in groundwater (GW) estimates derived from Gravity Recovery and Climate Experiment (GRACE) datasets are explored across the CONUS. Here, we seek to better understand how SW model selection affects GRACE GW uncertainty over the conterminous United States (CONUS) using 50 unique simulations from 2004 to 2010. Analysis of storage contribution to GW trend shows that where trend in TWS is weak or neligible, SW models can introduce new, uncertain trends into simulated GW. Where trend in TWS is strong, SW models increase the uncertainty in GW trend spatial extents. We find moderately higher levels of GW trend uncertainty than previous studies in the southwest CONUS (9-21%) and note exceptionally high uncertainty in the northwest and east CONUS (45-54%) attributed primarily to disagreement on trend direction between modeled soil moisture estimates. Results suggest that GRACE groundwater trends are obscured by SW model uncertainty in regions of low-magnitude TWS change, requiring satellite observations to be heavily supplemented with in-situ data to support GRACE as an alternative observational system. Finally, this dissertation develops machine learning models to bridge the divide between coarse resolution hydrologic models and fine resolution remote sensing models. This study trained random forest and linear regression models to predict annual 30-m ET from the Operation Simplified Surface Energy Balance (SSEBop) model over 2 million km2 of the conterminous United States (CONUS) from 2010-2017. To produce models capable of ET prediction outside the timeframe of satellite data availability, input features were restricted to common climate, landcover, and geophysical datasets that are independent from RS observations. The random forest model universally performed best, with average normalized root mean squared error of 19% over the CONUS, 10% over forested land in the eastern US, and 15% over forested land in the western US. Errors in western sites were disproportionately impacted by extensive shrub vegetation and barren surfaces where the model struggled to accurately capture SSEBop ET. The random forest model developed here allows for future use cases such as hindcasting and forecasting 30-m ET, evaluating the effects of land cover change on ET and basin hydrology at a variety of scales, and rapid fine-resolution ET approximation outside the CONUS. Results from this dissertation provide the scientific community with a comprehensive uncertainty analysis of modeled products at the continental scale and puts the effects of this uncertainty into practical application, allowing for insight regarding the focus of future modeling efforts. The use of machine learning models is shown to assist in bridging the divide between low resolution hydrologic models and fine resolution remote sensing models, offering the research community a path towards hindcasting and forecast the terrestrial hydrologic cycle at high resolution.