Urban microclimate modeling helps to quantitatively understand the fine level urban heat exposure distribution, which would provide an important reference for urban heat management. Fine level continuous sky view factor (SVF) maps are usually needed for modeling how the solar radiation fluxes reach the urban surface and impact the urban microclimate dynamics. However, the estimation of continuous SVF maps is very time-consuming, which limits the urban microclimate modeling to small geographical areas. In this study, we proposed to use graphics processing unit (GPU) parallel computing to accelerate the computing of SVF. The high-resolution digital surface models were used as the input to generate the continuous SVF maps of Philadelphia, PA, USA. Results show that the GPU-accelerated algorithms can reduce the time consumption of SVF computing by more than 99% using one NVIDIA GPU. Based on the computed SVF maps, this study further calculated and examined the spatial distribution of mean radiation temperature (T_mrt) as an indicator of outdoor heat exposure using the SOlar and LongWave Environmental Irradiance Geometry (SOLWEIG) model. There is an uneven distribution of the outdoor heat exposure in Philadelphia, and the eastern and southern parts have a higher heat exposure level. Increasing the tree canopy cover is an effective measure to reduce the heat exposure by providing more shade, especially for the eastern and southern parts of Philadelphia where shade is needed most.

城市小气候建模有助于定量了解城市热暴露的精细水平分布，为城市热管理提供重要参考。通常需要精细级别的连续天空视图因子 (SVF) 地图来模拟太阳辐射通量如何到达城市表面并影响城市小气候动态。然而，连续 SVF 地图的估计非常耗时，这将城市小气候建模限制在较小的地理区域。在这项研究中，我们提出使用图形处理单元（GPU）并行计算来加速 SVF 的计算。高分辨率数字表面模型用作输入以生成美国宾夕法尼亚州费城的连续 SVF 地图。结果表明，GPU 加速算法使用 1 个 NVIDIA GPU 可以将 SVF 计算的时间消耗减少 99% 以上。基于计算得到的SVF图，本研究进一步计算和检验了平均辐射温度的空间分布（T _mrt ) 作为使用太阳能和长波环境辐照度几何 (SOLWEIG) 模型的室外热暴露指标。费城室外热暴露分布不均，东部和南部的热暴露水平较高。增加树冠覆盖是通过提供更多遮荫来减少热暴露的有效措施，特别是对于最需要遮荫的费城东部和南部地区。