The urban signalized road network, characterized by its dynamic and complex nature due to frequent signal control adjustments and unpredictable demand fluctuations, presents significant challenges for predicting lane-level traffic flow. This study introduces the innovative MGCN-TAMA model, which addresses these challenges by integrating multi-graph convolutional networks with a temporal-aware multi-head attention mechanism. The proposed model employs three types of adjacency matrices-a geographical matrix, a signal matrix, and an attention matrix-to capture the complex spatial dependencies among various traffic approaches. Additionally, the model utilizes temporal-aware multi-head attention to discern the nonlinear correlations in traffic variations over time. Tested on a real-world dataset from Tongxiang City, the MGCN-TAMA model significantly outperforms traditional models. Notably, in the first 30-minute prediction interval, our model achieves the lowest Mean Absolute Error, with 2.5649 vehicles per 5-minute span. These results underscore the effectiveness of combining graph-based methods with advanced attention mechanisms to enhance the accuracy of predicting lane-level traffic volumes in urban networks.

The complex precipitation microphysics associated with super typhoon Lekima (2019) and its potential impacts on the consistency of multi-source datasets and radar quantitative precipitation estimation were disentangled using a suite of in situ and remote sensing observations around the waterlogged area in the groove windward slope (GWS) of Yandang Mountain (YDM) and Kuocang Mountain, China. The main findings include the following: (i) the quality control processing for radar and disdrometers, which collect raindrop size distribution (DSD) data, effectively enhances the self-consistency between radar measurements, such as radar reflectivity (ZH), differential reflectivity (ZDR), and the specific differential phase (KDP), as well as the consistency between radar, disdrometers, and gauges. (ii) The microphysical processes, in which breakup overwhelms coalescence in the coalescence-breakup balance of precipitation particles, noticeably make radar measurements prone to be breakup-dominated in radar volume gates, which accounts for the phenomenon where the high number concentration rather than the large size of drops contributes more to a given attenuation-corrected ZH (ZHC) and the significant deviation of attenuation-corrected ZDR (ZDRC) from its expected values (ẐDR) estimated by DSD-simulated ZDR-ZH relationships. (iii) The twin-parameter radar rainfall estimates based on measured ZH (ZHM) and ZDR (ZDRM), and their corrected counterparts ZHC and ZDRC, i.e., R(ZHM, ZDRM) and R(ZHC, ZDRC), both tend to overestimate rainfall around the GWS of YDM, mainly ascribed to the unique microphysical process in which the breakup-dominated small-sized drops above transition to the coalescence-dominated large-sized drops falling near the surface. (iv) The improved performance of R(ZHC, ẐDR) is attributed to the utilization of ẐDR, which equals physically converting breakup-dominated measurements in radar volume gates to their coalescence-dominated counterparts, and this also benefits from the better self-consistency between ZHC, Z¯DR, and KDP, as well as their consistency with the surface counterparts.