Remote structural characterization of thousands of buildings from San Jose, Costa Rica

Abstract

The present work shows the implementation of a data collection strategy for characterizing large amounts of buildings efficiently by the conduction of remote surveys on 360° panoramic images and aerial photographs. A set of 7,296 buildings from the Latin American city of San José, Costa Rica were studied and characterized from a structural engineering point of view, obtaining information like occupancy type, height, type of lateral load resisting system, and structural irregularities, among others. Also, an estimation of the error of the remote surveys was performed, by contrasting its results with the ones of field (in situ) surveys applied on a subset of 556 structures denominated “control buildings.” The results show that for San José buildings, the predominant occupancy type, height, type, and material of the lateral load resisting system are, respectively, residential, one or two storey, wall type of confined-reinforced masonry. The overall precision level estimated for the remote surveys was 75 %, which the authors consider acceptable and an improvement when compared to more popular surveys, for example, the field surveys carried out during a population and housing census that typically have an estimated precision level of 50 %. The results proved the adopted strategy to be a promising one, albeit subject to improvements to increase its precision and reduce the implementation time.

Impacto del modelo de consecuencias en la estimación del riesgo sísmico físico determinista para el cantón de San José, Costa Rica

“…Los resultados de la simulación numérica de la ruptura, propagación e identificación de la intensidad sísmica sobre los centroides de los barrios en el cantón central de San José se muestran en la Figura 5. El modelo de exposición considerado fue realizado por el Laboratorio de Ingeniería Sísmica (LIS) entre el año 2017 y 2019 y publicado por Esquivel-Salas et al [12]. Este cuenta con un inventario de 64807 estructuras generadas a partir de una muestra de 7294 edificios ubicados en el cantón de San José, Costa Rica.…”

“…Además, se incluye información de cantidad de niveles, material, sistema estructural ante cargas laterales, regularidad, propiedades de techo y piso, así como: ubicación geográfica, área de la huella, fecha aproximada de construcción y el distrito y barrio en el que se ubica. Dicha caracterización se llevó a cabo basándose en imágenes panorámicas de los edificios [12].…”

“…Para el cantón de San José se generan tres niveles de clasificación (taxonomía base agrupada, taxonomía base y taxonomía) de las estructuras adoptando las tipologías identificadas por Esquivel-Salas [12], lo que permitió además desglosar las pérdidas por los distintos niveles de clasificación.…”

Automatic Building Height Estimation: Machine Learning Models for Urban Image Analysis

“…The determination of seismic exposure is based on the characterisation of the attributes of the structure with different levels of detail to define the class of exposure of each building [7], as in the methods proposed by the Global Earthquake Model (GEM) foundation [4]. Previous research identified that one of the most influential attributes for seismic exposure classification is the height of the building or the number of floors [7,8]. As the attribute that is needed is the number of floors, a building height estimation with a maximum absolute error of one floor would be sufficient.…”

“…Research on the seismic vulnerability of buildings using imagery to extract basic parameters such as height or facade material is increasing worldwide [7]. Surveys are traditionally performed manually through visual inspection, which is time consuming and expensive.…”