Current land surface models, which are part of Earth System models, lack representation of coastal systems. Meanwhile coastal and estuary ecosystems can sequester relatively large amount of carbon per area in relation to other ecosystems. We are building now model of coastal area which consists from salt marshes, freshwater wetlands and grassland on example of River Don Delta. Two state-of-the-art models are being modified and combined for representation of the delta: three-dimensional hydrological model EX-MARE and vegetation distribution and carbon cycle model SEVER dynamic global vegetation model (DGVM). DGVM is being extended by implementation of a new plant functional type (PFT): salt marshes and by inclusion new PFTs traits. These traits are salinity tolerance, new allocation scheme between leaves and roots and height restriction implemented to describe a reaction of plants for inundation caused by periodic large marine surges. Hydrological model provides DGVM with depth of water in an area and amount of sediments which can be utilized for accretion and further change in plant heights. Important feature of combined delta modeling framework is process-oriented model of fires, which are frequent in the Don Delta. It was found in a preliminary statistical analysis that spatial distribution of frequency of floods is related to the frequency and intensity of fires, which are now being implemented into DGVM model. We are now trying to estimate possible carbon sequestration and fluxes and vegetation redistribution in the Delta in conditions of more arid climate, changes in large marine surges frequency and general shortage of fluvial sediments observed in the last 40 years. Remote sensing data and our field measurements are used for calibration and validation of the modeling framework.