Construction of urban turbulent flow database with wavelet-based compression: A study with large-eddy simulation of flow and dispersion in block-arrayed building group model

Urban Turbulent flow database, which includes accurately simulated velocity field with full spatial distribution and time-dependent dynamics, is beneficial for fast analysis of pollutant dispersion and comprehensive validation. However, the enormous data volume makes storage and sharing difficult. This research investigated the feasibility of constructing a small-sized database by compression. A wavelet-based compression method was selected because of its light calculation burden and independence of time series data. The database of the flow field in urban-like arrayed blocks was obtained by large-eddy simulation, and then the velocity and diffusion coefficient fields were compressed with different error controls to construct databases. The compression ability of the wavelet-based method was evaluated, and it was found that the maximum compression ratio is reached when most information in the wavelet coefficients is discarded. The effects of compression error on both the single snapshot and the time-series results were studied. The applicability of the compressed database was verified by the resimulation of passive scalar dispersion in the flow. According to results, the database with approximately 100-times compression is suitable for such applications as the visualization and dispersion simulation of passive scalars with sufficient accuracy. •An urban turbulent flow database with lightweight data size was constructed.•LES of flow and dispersion in a block-arrayed building group model was conducted.•Velocity and diffusion coefficient fields were compressed with a wavelet-based method.•Data size reduction of approximately 1/100 was achieved, maintaining sufficient accuracy.•Simulation of dispersion using compressed data showed sufficient prediction accuracy.