The COVID-19 pandemic is changing the way we work and play¹. As more people work remotely from home, visits to local parks have surged as people look for ways to stay active, find solace and quench their thirst for adventure². Such opportunities for leisure and recreation play an important part in people’s physical and psychological well-being³.

An elderly man sketching the landscape at Bishan-Ang Mo Kio Park in Singapore

Not all parks are created equal. For example, people seeking solace may be drawn to waterfront or nature parks with open or tranquil landscapes⁴. For others who love adventure (like myself), bike trails are great to break a sweat while having fun with friends. But most of the bike trails in Singapore⁵ are really hard to get to from where I live (not to mention figuring out bike transport). So, instead of heading out for some adventure, I tend to visit the open field and playground closest to home, going out for short walks with my son (still waiting for my next adrenaline fix!).

Waterfront landscapes at Upper Peirce Reservoir in Singapore

A very different kind of ‘waterfront’ closer to home

In city planning, park provision is typically measured by summarising the park area within a given region⁶. Beyond the area of parks, however, there is a need to characterise the wide variety of parks that serve different groups of people. Understanding such nuances in park access can benefit both residents and city planners, for instance, when considering the location of a new home, or when planning for new parks in the city.

Basic classification of parks in Singapore based on OpenStreetMap (OSM) data in 2020. Includes outdoor spaces such as beaches and informal nature areas.

The new R package home2park provides ways to measure a variety of park features related to recreation such as foot/cycle trails, waterfronts, forests, open spaces, playgrounds and sport/fitness amenities. It includes functions to download such data from OSM. Alternatively, you may supply your own proprietary datasets, or use new data for the purpose of future scenario planning. Summarising these features at each park allows us to make quantitative comparisons between parks. Some examples are shown below for the city of Singapore, based on OSM data in the year 2020. For instance, parks with extensive trails provide opportunities for running/wheeled sports:

Length of trails (foot- and cycle-ways) in Singapore

Cycling with our little one along East Coast Park (a rather rare occasion)

Other features such as water and vegetation are associated with visual relief, as well as restorative effects on people’s psychological well-being⁷. If such data (e.g. satellite imagery) are available, they can also be summarised at each park:

Length of waterfronts within & near parks in Singapore

Area of dense vegetation within parks in Singapore

Large water bodies and forest patches are present at Upper Peirce Reservoir Park

To calculate the ‘supply’ of these parks features to homes city-wide, these summarised values per park can then be assigned to each residential building. Since people are less likely to visit parks further away from their homes, a ‘distance decay’ parameter can be included. This reduces the supply value originating from parks further away. The ‘distance decay’ for walking to urban parks was empirically determined in Tu et al.⁸ to fit a negative exponential curve with a coefficient c value of 0.661 (see home2park package for details). The example below shows the supply of park area to residential buildings in Singapore.

The supply of park area to residential buildings in the year 2020, based on a ‘distance decay’ coefficient c value of 0.661. Each building is denoted as a point (a random subset is shown). Details in the package website.

Notice how the largest parks in Singapore tend to be centrally located, especially since we include informal nature areas (e.g. Central Catchment Nature Reserve). Buildings closer to these huge parks would thus have a much larger ‘supply’ of park area. However, each of the residential buildings would have different numbers of people living in them. A large park next to high-rise apartments would benefit a lot more people compared to low-rise housing. If we take into account the number of residents living within each building, a very different picture emerges!

The supply of park area to all building residents in the year 2020. Each building is denoted as a point (a random subset is shown). Details in the package website.

These building-level metrics can subsequently be summarised across larger regions, for example, based on zones used in city planning. Not surprisingly, the supply metric for park area derived from individual buildings is vastly different from conventional metrics used today (examples in Tan and Samsudin⁹). Feel free to explore these metrics for the provision of park area in the interactive map below.

In these unprecedented times, significant changes to human mobility and working environments will require us to rethink land use distribution in the city, particularly for office, residential and public spaces such as parks. Some evidence also point to shifting preferences, for example, toward wilder green areas¹⁰. These issues are starting to become a part of the national conversation in Singapore ¹¹, which is important considering that the city has major plans for sustainable development and greening over the next decade¹². As part of the Singapore Green Plan 2030, there will be 1,000 ha (more than 1,800 football fields!) more parks and park connectors, and every household will live just a 10-minute walk away from a park¹³. In view of these major plans, there is a need to move beyond basic summaries of area, and consider the variety of parks features that are important for outdoor recreation. The R package home2park represents a part of our effort to contribute to a more nuanced understanding of outdoor recreation and its spatial provision in cities. Note that the package is still experimental, so do reach out if you would like to contribute to improvements, or to report any bugs.