Dissolved organic carbon (DOC), a significant component of carbon budget, is of vital importance to aquatic biogeochemistry. Soil organic carbon (SOC) is known as the major source of DOC. Although significant advancement has been made in large-scale modeling of SOC modeling, modeling of SOC conversion to DOC, hence riverine DOC modeling, is still limited by lack of observational constraints on essential parameters controlling the fate and transport of SOC and DOC, particularly the solubility coefficient of SOC, Ks. To address this gap, we develop a predictive model of Ks applicable over the whole U.S. by combining artificial intelligence techniques and over 1000 observed Ks values (derived from in-situ observational data) over the U.S. We then generate a map of Ks over the contiguous U.S. in a vector format with respect to over 2.6 million river segments from the National Hydrography Dataset Plus (NHDPlus) dataset. Finally, we cross-validate the optimized model using observed Ks values from an additional set of USGS sites. This map lays a critical cornerstone for large-scale riverine carbon modeling for improving understanding of carbon budget at the regional and larger scales.