Knowledge of the renewable water resources of a watershed is strategic information which is vital for the long term planning of water and food security. In this study, we used a Soil and Water Assessment Tool (SWAT) model in combination with the sequential uncertainty fitting algorithm (SUFI–2) to simulate the water resource components (blue and green water) in the data scarce Kohnak watershed (in southwest Iran) based on river discharges. The simulation was performed for the period from 1992 to 2009 by considering the first three years as warm up. Due to imperfect incomplete climate data, two solution methods,(1) combining CRU data with observed climate data, and (2) integrating expert knowledge in defining uncertainty parameter ranges in the calibrating period, were used to increase the model accuracy prediction. Sensitivity and uncertainty analyses were also performed to improve the model performance. Simulated water resources components of blue water flow, green water storage, and green water flow were evaluated at the subbasin scale. The results showed that with the applied solution methods, SWAT could satisfactorily predict discharge flows and water resource components in the Kohnak watershed. The spatial variabilities of blue and green waters were a function of the spatial variability of precipitation, soil depth, land cover type and slope. Both the blue and green water flows decreased from upstream to downstream. The green water storage was larger in the middle and lower subbasin. It indicates that these regions have relatively sufficient precipitation and green water resources, which are beneficial for the development of rain fed agriculture. This study showed that in data scarce watersheds, a SWAT model in combination with expert knowledge can be used as a suitable tool for water component prediction.