Urban Park Use in the United States at Census Tract 2017 - 2021

We estimated seasonal park usage for 69,450 census tracts in the contiguous US from 2017 - 2021. We defined realized park usage as the proportion of the population of a given census tract that visits a park at least once in a given week. We estimated this quantity using aggregated and anonymized data from Google users who have enabled the Location History setting, which is optional and off by default. We estimated park usage per census tract between December 2017 and November 2021 as follows. First, for the purpose of this research, we estimated the census tract of residence of each Google user each month based on where that user spent most of their time, including nights and weekends. Second, for each week, we calculated the proportion of Google users residing in a given census tract who visited any park at least once that week. Park visits lasting less than 10 minutes or greater than 4 hours were excluded; very long visits are excluded as they may be for reasons other than recreation or reflect geolocation error among those that live or work very close to a park. Note that data on the mode of transport to or from a park were not available. Thus, park visits are counted regardless of the mode of transport used. Finally, we averaged weekly estimates to obtain seasonal averages of park usage, with seasons defined based on calendar months: winter (December, January, February), spring (March, April, May), summer (June, July, August), and fall (September, October, November). This computation uses differential privacy to protect user information.

我们针对美国本土48州的69450个人口普查区（census tract）估算了2017至2021年间的公园季节性使用情况。本研究将实际公园使用量定义为：指定人口普查区内，在给定一周内至少到访公园一次的居民占该区域总人口的比例。我们通过启用了位置历史设置（该功能为可选配置，默认处于关闭状态）的谷歌用户的聚合匿名数据集来估算该指标。
我们按如下步骤估算了2017年12月至2021年11月间各人口普查区的公园使用量：第一，为本研究目的，我们基于每位谷歌用户每月的日常活动主要聚集区域（包括夜间与周末时段），估算其每月的居住人口普查区；第二，针对每一周，我们计算居住在指定人口普查区内的谷歌用户中，当周至少到访过任意一处公园的用户占比。其中，停留时长不足10分钟或超过4小时的公园到访将被排除：过长停留可能并非出于休闲娱乐目的，或反映了居住或工作地点紧邻公园的用户的地理定位误差。需说明的是，本研究未获取往返公园的交通方式相关数据，因此公园到访统计不区分所采用的交通方式。最后，我们对周度估算结果取平均值，以得到公园使用量的季节性均值，季节划分依据自然月份：冬季（12月、1月、2月）、春季（3月、4月、5月）、夏季（6月、7月、8月）、秋季（9月、10月、11月）。本计算采用差分隐私（differential privacy）技术以保护用户个人信息。