A rapid and scalable radiation transfer model for complex urban domains

Abstract

An important component of urban microclimate is the radiative heat transfer between the myriad elements that make up the urban fabric. While great progress has been made in developing radiation models for idealized urban spaces, operational simulations of fully resolved city-scale domains remain elusive. As a result, simplifications and assumptions must be made. Such compromises may limit utility, and reveal a need for new, scalable microclimate models. This paper presents a novel, physically-based, building-resolving radiation transfer model (QESRadiant) that utilizes ray tracing techniques accelerated with graphics processing units (GPUs). QESRadiant builds on computer graphics methods, incorporating approaches for global illumination and light transport. Tests show that our methods can rapidly simulate the radiation balance for millions of surfaces with unique shapes and properties using a single consumer-class workstation on the order of several minutes. High-resolution (0.5 m) street canyon radiation budgets are validated using field data covering nine months. The results show the model is able to predict radiative fluxes in the canyon with overall average R 2 of 0.77 and mean error of 14.5 Wm −2. Solar radiation was extremely sensitive to geometric obstructions, demonstrating the need for high-quality field experiments that resolve urban details affecting the radiation balance.

The Microscale Urban Surface Energy (MUSE) Model for Real Urban Application

“…However, they do not apply to complicated morphologies in real urban environments. Some previous studies have used ray tracing algorithms (e.g., [40,42,43]). When a large number of photons tracks in the algorithm, it can calculate the view factors with high computation accuracy [61,62].…”

“…The tower measurements are used to provide accurate meteorological forcing for validation of the MUSE model. The global shortwave and longwave radiative fluxes (S dir↓ , S di f ↓ , and L di f ↓ ) from a four-component radiometer (CNR4, Campbell Scientific Inc., Cache Valley, UT, USA) deployed at the meteorological tower are applied homogeneously to the active patches, in which the measured global shortwave radiative flux is partitioned to the direct and the diffuse radiative fluxes following the empirical method [43,71]. The wind speed (U a ) and air temperature (T a ) adjacent to the active patches are obtained with interpolation from the wind speed and air temperature profiles computed using the flux measurements from the three-dimensional sonic anemometer (CSAT3, Campbell Scientific Inc., USA) and the Monin-Obukhov similarity relations as follows:…”

“…Therefore, the empirical corrections of the flat surface albedos were applied for validation purposes. The downward shortwave radiation and its partitioning into the direct and diffuse shortwave radiation are calculated based on empirical formulas in Overby et al [43] with clear-day atmospheric turbidity conditions in Monteith and Unsworth [71].…”

A reverse ray-tracing method for modelling the net radiative flux in leaf-resolving plant canopy simulations

“…In order to ensure that rays cover the entire footprint of the domain, a "bounding sphere" can be constructed that encompasses all modelled objects. Rays are then cast from a disk with the same diameter as the bounding sphere, that is positioned tangentially to the bounding sphere (see Overby et al, 2016). The energy assigned to each ray is Q s πR 2 s /N rays , where Q s is the radiative flux normal to the source direction, R s is the radius of the domain bounding sphere, and N rays is the number of direct rays launched toward the domain.…”

Coupled fire-atmosphere-smoke forecasting: current capabilities and plans for the future

“…The radiation and convective heat transfer from the fire to the atmosphere, the buoyancy effects on the fire-line winds, flow over complex terrain, changing fuel types, and structures (including canopies or individual trees and buildings) must be accounted for while remaining computationally viable. The ongoing work focuses on possible solutions in the form of coupling with complementary, reduced-physics diagnostic models such as QUIC-EnvSim (Overby et al, 2016). In this approach, the WRF model would be a source of coarser-scale weather data used to drive the prognostic QUIC model similarly as it was done for urban simulations presented in Kochanski et al (2015).…”