Efficient matrix algebra encoding for urban solar irradiation simulation: fine-grid ground-level estimation with vector data

Abstract

Conducting a detailed assessment of solar irradiation at the pedestrian scale on all ground surfaces of a city can assist in identifying cooler routes for pedestrian navigation or preparing the city for potential overheating issues by pinpointing overexposed areas. This article proposes an effective method for conducting such an assessment within a GIS with metric resolution across territories exceeding 100 km². It is based on standard datasets and implements an efficient strategy that relies on separation of variables, domain decomposition, and dimensionality reduction. This strategy involves creating a synthetic representation of the facades of the surrounding buildings (spatial dimension) which accelerates the calculation of shadows based on the sun’s position (temporal dimension). To demonstrate the effectiveness of this method, we applied it to a French city, generating fourteen maps illustrating the solar irradiation of the area for different months of the year or for two given dates with specific weather conditions. The proposed strategy, along with the synthetic representation of building facades, opens up a wide range of possibilities. In addition to synthesizing machine learning labeled datasets, we can also consider calculating solar irradiation with time steps of a few minutes to update weather conditions throughout a journey.