This dissertation is devoted to the improved observation, understanding, and modeling of the global hydrological cycle through the integration of data from satellite remote sensing. Recent advancements in retrieval algorithms and the maturation of numerous satellite missions have made available an unprecedented record of land surface and atmospheric profile observations at coarse (100 km) to moderate (1 km) spatial resolution and sub-daily temporal repeat. For complex processes that occur over multiple spatial and temporal scales, such as land-atmosphere feedbacks (coupling), satellite remote sensing offers the only feasible means of observation. Retrievals of near-surface temperature and humidity from the Atmospheric Infrared Sounder (AIRS) form the basis for much of the research completed herein. They are first validated against model forcing fields and observations from over 2,000 ground stations across the continental USA and Africa and then used as inputs to a modified Penman-Monteith evapotranspiration (E) model, forced only with remote-sensing data. Sensitivity of the estimates of E to errors in inputs from AIRS and various other satellite-based meteorological and vegetation datasets is quantified using a 24-member ensemble of simulations. Next, AIRS-derived estimates of low-level humidity index and convective triggering potential, validated at over 450 radiosonde stations worldwide, are used to produce the first global, observation-based map of predominant coupling signals. Last, a global map of AIRS-derived lifting condensation level and Advanced Microwave Scanning Radiometer soil moisture correlation strength, a diagnostic metric for coupling, is inter-compared with correlations derived from several land surface models and reanalysis. Overall, the results highlight the value of satellite-based data sets for land surface hydrology and coupling studies, but also exhibit clearly the inherent uncertainties of their application. The global maps of coupling produced in this dissertation represent a significant contribution to the improved understanding of coupling. By successfully identifying regions where coupling persists, they lead the way for future detailed process and event-scale observational (on-the-ground, model, and satellite-based) studies. Initial findings suggest that real world coupling is less pronounced than that maintained by most models.