Modern traffic control and management systems in urban networks require real-time estimation of the traffic states. In this paper, a novel approach for modeling traffic flow in urban networks that is especially suitable for state estimation is proposed. The complexity of the urban traffic model is reduced by assuming availability of connected vehicle data. We first investigate the observability issue in urban traffic networks using a graphical approach. Then, the proposed model for the evolution of the traffic flow in urban traffic networks is developed and used in two layers, i.e., link layer and network layer, to estimate, in high-resolution (second-by-second), the traffic states in the whole network. Traffic states in the link layer include queue tail location and the number of vehicles in the queue, while in the network layer, estimation of the total number of vehicles per link and turning rates at the intersections is carried out. In a first step, it is shown that the estimation approach only requires the detectors at the borders of the network. We further demonstrate that in the proposed scheme, one may reduce or drop the need for spot detectors for the price of reduced, but still reasonable estimation accuracy. The validation of the approach has been undertaken by comparing the produced estimates with realistic micro-simulation results as ground truth, and the achieved simulation results are promising.