Levantamento e modelagem da vegetação arbórea para simulação computacional da iluminação natural

Abstract

The arboreal vegetation is an element of difficult insertion in the prediction of daylighting. To simulate an environment with trees, in daylight simulation software, it is necessary to develop three-dimensional models that represent all the elements intervening in the reproduction of light. Ideally, variables such as height, crown geometry, quantity, leaf size and spacing, trunk size, branches and twigs, reflectance indices, opacity indices, among other variables, should be reproduced in the three-dimensional model. In addition, the complex modeling also depends on a thorough survey of all the variables listed. Thus, this thesis has the main objective of characterizing a simplified and reproducible method for surveying and modeling tree vegetation in the computer simulation of daylight. The data for this research comes from two sources: measurements under real sky conditions and computer simulations of daylight. Illuminance and luminance records (HDR images, or High Dynamic Range images) make up the research database. Comparative analyses were performed to validate the results of the computer simulations, verifying to what extent the 3D computer model of the tree adequately represents the tree used in the field collection (real sky). The main comparative results between measurements and simulations indicated strong to moderate correlations (Spearman Correlation) in all scenarios studied. It was observed that the errors found from the statistical indicators MBE (Mean Bias Error) and RMSE (Root Mean Squarred Error) tend to be larger, especially at the points closest to the window, where they are exposed to a greater amount and variability of light. This result means that there is a greater difficulty in predicting illuminance more accurately in locations with excessive illuminance. It was observed that statistically there is a correlation between the DGP (Discomfort Glare Probability) extracted from the HDR images and the DGP extracted from the simulations, but weak. The coefficient of determination (R²) indicated that the results are statistically significant in 47% of the observations. These results were already expected, mainly due to the direct entrance of sunlight into the space at some times, and the consequent alteration of the illuminance distribution on the surfaces. In general, it could be assumed that the main sources of discrepancies between the measurements performed under real sky and the computational simulations were: 1) absence of the surroundings in the computational model; 2) differences between sky types (real sky and the Perez sky model used in the simulations); 3) the presence of direct sunlight; 4) differences in the photometric characteristics of the materials; 5) scale of the devices used in the field measurements. Differences between measurement and simulation results are expected in simulation practice due to the simplifications of the light phenomena occurring, especially in the case of computer simulations involving elements as complex as tree canopies. Thus, it is concluded that the method employed in this study can contribute to the simulation of canopy permeability to the passage of natural light.