NirmitSachde/carbon-clock-data

该数据集名为Carbon Clock — Manhattan · Trip data，用于支持The Carbon Clock — Manhattan项目，这是一个24小时动画可视化，展示纽约市交通、排放和空气质量。数据集包含一个二进制文件trips-2m.bin，记录了2025年1月在曼哈顿开始和结束的2,933,898次黄色出租车行程。数据来源于NYC TLC黄色出租车行程记录，其中2,933,750次行程通过OSRM v5在OpenStreetMap道路上路由成功（成功率99.995%）。几何形状使用Douglas–Peucker算法简化（容差30米），保留所有转弯点并去除共线噪声。总路点数为20,489,760个，平均每次行程7.0个路点。格式为自定义二进制CCM1，包含Uint32偏移表、每行程的RGBA颜色以及每个路点的Float32 [经度, 纬度, 自午夜以来的秒数]，相比等效JSON文件大小减少约40%，解析速度提高约10倍，可直接用于Deck.gl的TripsLayer作为GPU数据缓冲区。文件大小约257 MB，通过Hugging Face托管以支持前端跨域获取，无需代理。许可证方面，二进制格式和处理管道为MIT许可，底层出租车数据来自NYC TLC，属于公共领域。This dataset, named Carbon Clock — Manhattan · Trip data, supports The Carbon Clock — Manhattan project, a 24-hour animated visualization of NYC traffic, emissions, and air quality. It includes a binary file trips-2m.bin containing 2,933,898 yellow-taxi trips that started and ended in Manhattan during January 2025. The data is sourced from NYC TLC Yellow Taxi Trip Records, with 2,933,750 trips successfully routed through OpenStreetMap roads via OSRM v5 (99.995% success rate). Geometry is simplified using the Douglas–Peucker algorithm (30 m tolerance), preserving every turn while dropping collinear noise. There are 20,489,760 total waypoints, averaging 7.0 per trip. The format is a custom binary CCM1, consisting of a Uint32 offset table, RGBA color per trip, and Float32 [longitude, latitude, seconds-since-midnight] per waypoint, reducing size by ~40% compared to equivalent JSON and parsing ~10× faster, enabling direct use as a Deck.gl TripsLayer GPU data buffer. The file size is approximately 257 MB, hosted on Hugging Face to enable cross-origin fetching for the frontend without a proxy. The license is MIT for the binary format and processing pipeline, while the underlying taxi data from NYC TLC remains in the public domain.