A global scale approach involving data assimilation schemes (e.g. Distributed Oceanographic Data System) is designed to simulate the discharge of sediment to the coastal ocean at the dynamic level (daily). The result is either a real-time, hindcast or forecast picture of coastal hydrology optimized to estimate sediment loads of rivers. The approach links a compendium of global and regional web-based databases into a GIS system. Relational and spatial methods (i.e. RiverToolst, HYDRO1k, ArcInfot) facilitate the process of data acquisition useful to sediment discharge models (i.e. HydroTrend). As a climate-driven hydrological model, HydroTrend incorporates drainage basin properties (river networks, hypsometry, relief, lakes or reservoirs, distributary channels) through high-resolution digital elevation models, along with other biophysical parameters (basin-average temperature, precipitation, evapo-transpiration, canopy, soil depth, hydraulic conductivity, ice fields). The schema is designed to provide important boundary conditions for marine sediment-dispersal models, concomitant with ocean data (wind, wave, currents). Considering that <4% of world-rivers are monitored for their sediment loads, the approach provides a unique means to predict the sediment flux across an entire coastline at a high-resolution temporal scale. Model comparison to long-term (1950–1994) observations from the Lanyang River (Hsi), Taiwan is shown to capture average conditions and inter- and intra-annual variability of water discharge, sediment concentration and loads.