Urban drivers frequently experience challenges posed by adverse weather conditions like heavy rain. Despite the influence of these conditions on individual travel behaviour, there is limited understanding of how various aspects of rainfall affect the route choices in urban trips. In this context, this paper aims to evaluate the impacts of different rainfall conditions on the route choice behaviour of private vehicle users. We propose a method to identify the specific routes taken by drivers, utilizing observed trips reproduced from a large-scale dataset containing license plate information recognized by speed device cameras. Path Size Multinomial Logit models penalizing routes that share common links were estimated based on a choice set comprised of routes with a maximum similarity threshold. Distance and the actual travel times in the routes were estimated using real-world data from a third-party company, and cumulative, average and maximum rainfall through each route of the choice set were obtained from data captured by meteorological radar. A case study in the city of São Paulo, Brazil, considered the average and maximum rainfall intensities, distance, and travel time in different models. The results indicate that both average and maximum rainfall intensities negatively impact the utility of routes. This research enables the identification of routes with the highest probabilities of being chosen during intense rainfall. Such information is valuable for implementing measures to minimize the impacts caused by adverse weather conditions in urban trips.

中文翻译：

---

降雨对私家车用户城市路线选择的影响：来自巴西圣保罗的见解


城市司机经常遇到大雨等恶劣天气条件带来的挑战。尽管这些条件对个人出行行为有影响，但人们对降雨的各个方面如何影响城市出行的路线选择的了解有限。在此背景下，本文旨在评估不同降雨条件对私家车用户路径选择行为的影响。我们提出了一种方法来识别驾驶员所采取的具体路线，利用从包含测速设备摄像头识别的车牌信息的大规模数据集中复制的观察行程。路径大小多项式 Logit 模型基于由具有最大相似性阈值的路由组成的选择集来估计共享公共链接的惩罚路由。使用第三方公司的真实数据估算路线中的距离和实际行驶时间，并从气象雷达捕获的数据中获得选择集每条路线的累积降雨量、平均降雨量和最大降雨量。巴西圣保罗市的一项案例研究考虑了不同模型中的平均和最大降雨强度、距离和旅行时间。结果表明，平均和最大降雨强度都会对路线的利用率产生负面影响。这项研究能够识别在强降雨期间最有可能被选择的路线。此类信息对于采取措施最大限度地减少城市出行中恶劣天气条件造成的影响非常有价值。