An IoT approach to characterising biodiversity of green spaces for planetary health
< Project overview >
Greater biodiversity in natural spaces could potentially provide restorative health benefits and offer indirect health benefits through enhanced ecosystem functions, disease regulation, and exposure to microbial diversity (Wood et al., 2019). However, capturing information on biodiversity is complicated and highly resource intensive, and little research exist to date on how local governments could incorporate biodiversity into planning and designing of green spaces to improve ‘planetary health,’ a term that describes the interconnection between human health and the health of the planet.
In this project, we aimed to explore IoT-enabled mechanisms that can be used to capture information on greenspace biodiversity specifically looking at sound-scape information. IoT-enabled mechanisms have the potential to significantly reduce the costs of data collection as well as leveraging citizen science, so allowing local governments to incorporate biodiversity into planning and design of green spaces.
In this project, partners from the University of Oxford, Oxfordshire County Council, Newcastle University and Open Acoustic Devices came together to explore opportunities to identify and draw together IoT-enabled solutions to positively impact this at scale.
Despite the promise around greenspace to improve health, robust evidence on the characteristics of greenspace that yield the best health outcomes is lacking, as highlighted by Barton and Rogerson. One of the challenges related to the above is capturing relevant information that can be used to understand the greenspace characteristics that deliver the best health outcomes.
This project built on the successful Greenspace Hack Pitch-In project to firstly understand the impact of biodiversity on health and secondly, to demonstrate how IoT can be used to capture additional greenspace characteristics on biodiversity to complement the information we captured using Greenspace Hack’s eNEST tool. As with Greenspace Hack, this information can be used by county councils and urban planners to enable citizens to use technology to inform the design and build of healthier places, in line with the vision of the NHS Healthy New Town initiative.
Through our project we anticipated to address the following knowledge exchange barriers:
Lack of knowledge of various stakeholder skills and interests.
Lack of understanding of the full landscape of possible architectures for a possible IoT solution.
Lack of understanding in how IoT will/can generate value in a given application domain.
Lack of knowledge/understanding of how analytical tools can be used for leveraging IoT data.
These barriers are of obvious importance because addressing them means we will have a better understanding of how IoT can benefit urban planning of greenspaces while also giving us an indication of how IoT-based solutions can be used in practice.
What was done
To explore our project aims, we:
Conducted a scoping literature review to understand the relationship between biodiversity and health – (manuscript under preparation for the Ecosystem Services Journal). We conducted a systematic database search and thorough screening of articles, conducting a review of five previous reviews and narrative synthesis of the ten recent studies which met our inclusion criteria. We also performed a bibliometric analysis of 1,758 studies to chart geographical, temporal and topical trends in the field.
Explored an IoT-based audio detection solution – we explored how AudioMoth, a validated low-cost, full-spectrum acoustic logger developed by Professor Alex Rogers at the University of Oxford’s Computer Science Department, can be utilised in a citizen-sensing approach to capture soundscape information of greenspaces in Oxfordshire and Newcastle. We then explored how this information could be used to extrapolate biodiversity information. We used the networks established in Oxfordshire as part of the Greenspace Hack project to recruit volunteers and also worked with partners in Newcastle to recruit volunteers.
Scoping literature review
The results of our literature review revealed that few reviews have holistically analysed the evidence for a relationship between biodiversity in greenspaces and human health directly. Those that did, found mixed, or weak evidence for a relationship between biodiversity and various aspects of physical and mental health.
Our narrative review discovered evidence supporting associations between health and floral biodiversity, particularly subjective wellbeing and self-reported health, with mixed evidence for other health outcomes or more holistic measures of biodiversity. Further research and new methods are required to understand long-term health impacts of exposure to biodiversity through larger-scale longitudinal and controlled case studies.
Biodiversity may be an important factor in health-supporting effects of greenspaces.
A bibliometric analysis revealed a small but growing international research field.
A review of reviews and narrative synthesis provided some limited evidence.
Floral biodiversity is associated with improved wellbeing and subjective health.
More research and larger studies consistently characterising biodiversity are needed.
Using an IoT-based solution to capture soundscape information and extrapolate biodiversity information
We partnered with the Oxfordshire County Council to recruit volunteers through an active engagement campaign through their networks:
Volunteer information was captured through an online survey form. Despite the difficulties with the COVID-19 restrictions, we were able to recruit 22 volunteers who committed to deploy the Audiomoth devices in compliance with COVID-19 social distancing regulations and following instructions posted on our Greenspace Hack website.
In addition to community volunteers, the Innovation team at Blenheim Estate also agreed to deploy sensors across selected locations within their 1,000 acres of formal gardens.
Data was organised into 14 geographical groups, with comparable environments, defined by deployed devices within 3km of each other, which are highlighted in the red boxes on the map below.
Devices within each group were usually deployed for about the same length of time and the timeline below shows the quantity of data collected:
NDSI and acoustic energy
We used two indices to analyse the acoustic environment:
Normalised Difference Soundscape Index: this produces values between -1.0 and +1.0 and represents ratio of biotic to anthropogenic noise and uses frequency bands to define each.
Acoustic energy: represents the level of noise within a recording, which can be used to identify times of peak activity at a site.
Data output case study – Blenheim Estate
Two clusters of devices were chosen within the group at Blenheim Estate –cluster one was deployed alongside the water and cluster two was deployed at the furthest point from the water of the deployment area:
Using our two indices, you can see a marked difference in both areas:
This data highlights important differences in the utility of these two indices and, based on the information captured, we decided that NDSI was much better for capturing rich information because of its ability to compare biotic vs non-biotic sounds.
Focusing on the NDSI for the area near vs far away from the waterside, we can see the NDSI ratio changing at different times of the day and also varying between the different locations – thus giving us a richer indication of the variation and extent of biotic noise.
Focussing on the periods where the NDSI ratio increases (ie more biotic noise), we found that there was a substantial increase in birdsong (audio data available upon request). Furthermore, we also found that the NDSI changed depending on whether the measurement was done on a weekday vs weekend:
This indicates that fauna change their patterns based on anthropogenic patterns – likely due to differences in traffic and rush hour.
Data output for all sites
Overall, NDSI seems to be a good metric for defining the ‘naturalness’ of a site. While acoustic energy is useful for studying the behaviour of an environment, it doesn’t have a huge effect on the naturalness. From the analyses of all of our sites, ranking by NDSI demonstrated that sensors deployed near Blenheim Palace ended up being the ‘most natural’.
One caveat is that it is difficult to judge the overall effectiveness of the NDSI metric due to lack of diversity in sites, which we will hopefully be able to overcome in future studies by exploring a greater variety of sites.
Throughout our project, we felt that we did a good job of addressing most of the barriers we initially set out to address.
Lack of knowledge of various stakeholder skills and interests:
We created an easy to follow campaign that supported a citizen science approach to the deployment of the AudioMoths, which included a simple set of instructions on set up of the device.
Lack of understanding of the full landscape of possible architectures for a possible IoT solution:
We were able to successfully use the AudioMoth IoT device for capturing soundscape information. That said, we felt we could have done a better job of trying to identify other IoT solutions to capture visual-scape information.
Lack of understanding in how IoT will/can generate value in a given application domain:
Through our strong partnership with the Oxfordshire County Council, we wanted to understand the utility of our approaches and particularly the NDSI for planning purposes. Through our work with the Oxfordshire County Council, we have been able to position ourselves to explore future projects where we can use the AudioMoth and NDSI as two low cost approaches that can to inform planning of greenspaces.
Lack of knowledge/understanding of how analytical tools can be used for leveraging IoT data:
Through our project, we utilised two methods (NDSI and acoustic energy) to analyse the soundscape information captured with the AudioMoths to glean information on biodiversity. Through our analyses, we feel confident that we can use the NDSI for our future work on greenspace designing and planning with the Oxfordshire County Council and local authorities more generally.
Deliverables and outputs
The scoping review is under preparation and will be submitted to the Ecosystem Services Journal.
For the AudioMoth data, we will be posting all of this data on the Greenspace Hack website so that it is publicly available. We will also be preparing an academic manuscript with the key findings of our project.
Mental health conditions are one of the most significant contributors of overall global disease burden and cost society an estimated £1.6 trillion per year. Greenspaces (maintained and unmaintained environmental areas including nature reserves, wilderness and urban parks) have been used in urban contexts for decades for aesthetic and recreational purposes and emerging evidence suggests that greenspace can improve mental health and wellbeing – indeed, individuals who use greenspace or engage in green exercise have less mental distress, less anxiety and depression and healthier cortisol levels.
Further to this, biodiversity of greenspaces is a specific characteristic that has been linked with the restorative benefits of greenspaces across different contexts and population demographics.
Despite the promise around greenspace to improve health, robust evidence on the characteristics of greenspace that yield the best health outcomes is lacking, as highlighted by Barton and Rogerson.
If greenspace were considered in the same way as a drug for mental health and well-being would be, more detailed understanding of its mechanisms would lead to optimal dosage, and knowledge of when and for whom it might work best. Optimal doses need to account for a wide range of mediators (Shanahan, Fuller, Bush, Lin and Gaston, 2015) including:
Environmental factors, both qualitative (eg biodiversity, air quality, noise) and quantitative (eg tree canopy cover), as well as weather.
Personal factors, such as age, gender, beliefs about the value of nature, nature relatedness, prior experiences and childhood memories, as well as perceptions of risk.
Social and community factors, including social interaction, trust, ethnic, cultural and social norms, and accessibility of green spaces.
One of the challenges related to the above is capturing relevant information that can be used to understand the greenspace characteristics that deliver the best health outcomes.
This project built on the successful Greenspace Hack Pitch-In project to demonstrate how IoT can be used to capture additional greenspace characteristics on biodiversity to complement the information we captured using Greenspace Hack’s eNEST tool. Moreover, the use of IoT and citizens provides a low-cost mechanism to capture this data, so offering a real data-collection mechanism for councils.
As with Greenspace Hack, this information can be used by county councils and urban planners to enable citizens to use technology to inform the design and build of healthier places, in line with the vision of the NHS Healthy New Town initiative. Indeed, the work of this project helped Oxfordshire County Council to include an IoT element, through the use of AudioMoth, in a NERC funded project OxAria 2.
Having established a method to extrapolate an IoT-informed measure of ‘naturalness’, we would now like to explore what this score actually means in practice.
We plan on deploying to a greater variety of sites to ensure the NDSI can be used across a range of settings. We also plan on applying for future funding to explore stakeholder engagement to link the ‘naturalness’ score and what we have learned from our scoping review to experiences of citizens of greenspaces.
Our goal would then be to help use this score to inform policy planning around greenspaces by Oxfordshire County Councils and local governments more generally.
The Pitch-In project was the catalyst in helping us bring together a diverse group of stakeholders committed to a common mission. The dedication of the team to the project was essential in helping up keep the momentum despite facing setbacks, the principle of which was adapting to the COVID-19 lockdown. While the lockdown prevented the County Council from deploying the sensors to council-run greenspaces, we were able to find a way around this by leveraging local community networks, and we were also introduced to Blenheim Estate, who became a key case study for us.
The key to the success of our partnership with Blenheim Estate was the strategic alignment of Nature Sensing to their innovation team’s existing work on IoT sensors and their use to monitor their 1,000+ acres of greenspace.
The COVID-19 lockdown was difficult to navigate. As a team, we could have mobilised faster to engage with our stakeholders more proactively so we could have collected our data more efficiently and to also capture more greenspaces to ensure the NDSI was representative of naturalness across a variety of different sites.
Some support on exploring other IoT options for capturing soundscape information would have been helpful. It also would have been great if we had some help to navigate IoT options to capture visual-scape information to identify flora/fauna aspects of biodiversity.
What has Pitch-In done for you
Pitch-In gave us an opportunity to bring together a diverse group of stakeholders committed to a common mission. Furthermore, because of the small scale and complex, high risk nature of our project (where we combined themes like IoT, citizen science, health and environmental health), Nature Sensing was the type of project that traditional funders would likely not have funded.
Pitch-In’s support served as a catalyst for us to build momentum for this essential work on identifying essential aspects of greenspaces that can promote health, which we will hopefully be able to continue developing over the coming months and years.
Dr Andy Hong – University of Oxford