Better understanding of the spatial and temporal distribution precipitation is critical to climatic, hydrologic, and ecological applications. However, lack of reliable precipitation observation in remote and developing regions poses a major challenge in the above applications. Recent development of satellite remote sensing techniques provides a unique option for better observation of precipitation from the limitation of ground measurement. This study presents the development of a satellite-based precipitation observation system based on artificial neural network (ANN) models, and their implementation for providing near global precipitation observation using combined geostationary and low orbital satellite imagery. Two major stages are involved in processing satellite image into surface rainfall rates. First, image texture and geometry features cloud longwave infrared image of geostationary satellites were extracted. Secondly, ANN models were used to classify image features to a number of cloud patterns and then process and associate those classified cloud image patterns to the surface rainfall rates. Model parameters were initially calibrated from gauge-adjusted radar rainfall. In addition, adaptive parameter adjustment from limited low-orbital satellite rainfall estimates were used, enabling the model to be extended to the remote regions and oceans. The system is operated to generate hourly rainfall from 50S-50N. Details of the system development, model validation, and error analysis of model estimates will be discussed. In addition, case studies of applying the ANN rainfall estimates to hydrologic modeling will be presented.