Development and Testing of a Spatially Resolved Urban Land Surface Model Utilizing Parallel Computing on Graphics Processing Units (GPUs)

The Green Environmental Urban Simulations for Sustainability (GEnUSiS) Project was organized to create a high-performance, building-resolving, small-scale meteorological simulation tool for urban environments. More specifically, the simulation tool is targeted at urban planning groups for use in urban design for reduced energy consumption, optimal pollutant dispersion, and improved human comfort through optimal land use such as green infrastructure. In order to provide faster than real-time results for city-size domains on a relatively inexpensive personal computer, the model has been optimized to make use of the parallel computing capability of graphics processing units (GPUs). To satisfy the project's goals, the temporal changes found throughout the atmospheric boundary layer (ABL) must be predicted in the model. These temporal changes are forced by surface fluxes of heat, moisture, and momentum, making the accurate prediction of surface fluxes extremely important for models attempting to correctly capture ABL physics. A land-surface model (LSM), which explicitly solves for the transport of heat and moisture through a ground column using a one-dimensional finite differencing approximation, is presented. The LSM is capable of modeling a variety of surfaces to include various soil types, walls and windows, and other surfaces such as impermeable roads and sidewalks. The LSM is coupled to a radiation model that utilizes the Nvidia OptiX ray tracing engine and a turbulence transport model that utilizes the Quick Urban Industrial Complex (QUIC) Dispersion Modeling System through surface energy and moisture budgets. The coupling utilizes an iterative technique that relies on Monin-Obukhov similarity theory for land-atmosphere fluxes. To validate the LSM, surface temperatures, sensible and latent heat fluxes, and sub-surface temperature profiles are compared to data from a series of test cases of increasing complexity. First, comparisons between the conduction portion of the LSM and the analytical conduction solution for both a building wall and ground case are presented. Next, a comparison between a homogenous atmospheric boundary layer test case and data obtained from the Mountain Terrain Atmospheric Modeling and Observations Program (MATERHORN) field experiment is presented. Finally, a comparison with the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) single building test case is presented.