Treated urban wastewater has been increasingly used for irrigation to increase crop production, but inadequate management of this marginal quality saline water cannot only reduce crop yield but also degrade soil. To increase sustainability, a sound irrigation policy based on real-time soil states is needed to avoid the adverse effects from excess salinity in soils. In real applications, uncertain model inputs and soil heterogeneity are expected when a modeling approach is used. In addition, sensor measurement information is generally insufficient to characterize the state of the soil profile in detail. To overcome these limitations, a monitoring and modeling system assimilating embedded sensor datastreams into a hydrologic model to improve real-time soil state estimates over the whole profile of interest is examined with a wastewater reuse system in Palmdale, CA. Experimental results show that under an unknown heterogeneous soil, the filtering scheme is able to describe the spatial variability of moisture and soil solution electrical conductivity in soils due to direct state estimation by assimilating moisture and electrical conductivity measurements and also due to improved soil hydraulic parameter estimates through the data assimilation framework. Two data withholding validation experiments were conducted to evaluate the added value of assimilating measurements. The spatial data withholding experiment demonstrates that state predictions at unobserved areas can be improved via the assimilation if the soil system is relatively homogeneous and effects from erroneous irrigation forcing are reduced. The temporal data withholding experiment indicates that a high update frequency can maintain the reliability of state estimation. Overall, this system can provide a more accurate soil characterization that could be used to optimize irrigation management to reduce adverse effects from wastewater reuse.