<p>Accurate traffic forecasting is vital for the efficient functioning of intelligent transportation systems, facilitating better traffic management and planning. However, existing studies face two major limitations: they fail to simultaneously capture spatial-temporal dependencies and can only predict traffic features for nodes with historical data, leaving nodes without historical data unaddressed. To overcome these limitations, we propose a GraphSAGE-based spatial-temporal synchronous traffic forecasting network considering sensorless roads (GSTSN), which offers two main advantages. First, GSTSN designs a spatial-temporal synchronous graph to integrate the representation of spatial and temporal correlations. Specifically, it forms a spatial compensatory graph by combining a spatial distance graph and a spatial similarity graph, which is then integrated with a temporal causal graph to create the spatial-temporal synchronous graph. Second, GSTSN employs GraphSAGE-based attention and mean aggregation functions tailored for nodes with and without historical data, enabling accurate traffic predictions for both types of nodes simultaneously. Extensive experiments on four real-world datasets, including scenarios with 10% and 20% sensor missing rates, show that GSTSN achieves superior performance compared to the state-of-the-art baselines, confirming its effectiveness and robustness.</p>