Workshop 2: Enriching Scotland’s soils data – sensing, modelling and interoperability
24 May 2022
Kelvin Hall, University of Glasgow
This workshop brought together researchers from fields including archaeology, heritage, agriculture, soil science, environmental science and engineering to focus on how we can enrich the soil records used in Scotland. Presentations focused on uses of sensing data, designing sensor networks, modelling from sensing data, and cross-domain data interoperability.
Workshop Aims
- to identify shared research and management priorities related to soil health and soil heritage in order to inform how we develop Scotland’s soils data
- to discuss the capability of commercial and experimental sensors, survey and sampling technologies, and connected analytical methods to provide data to address these shared priorities
- to identify barriers to the creation of interoperable remote and near-surface sensing data resources on soils
Key Themes
- Improving soil data in agricultural areas to enable environmentally sustainable land management practices
- Diversifying sources of data on soil, soil biomes, and crops, with an emphasis on remote and near surface sensing
- Breaking down data, methods and knowledge silos between agricultural, environmental and cultural and natural heritage domains
- Assessing data requirements for agri-environmental monitoring and management, and overlaps with cultural and natural heritage monitoring and management requirements
- Addressing problems introduced by data patchiness (irregular coverage spatially and temporally) and biases in distributions, which prevent scaling up of predictions and models
- Building the case for soil data as essential to sustainable rural communities
- Good practice for sensor and data acquisition and management protocols for natural environment scaled deployments
Discussion Prompts
Emerging from Discussion…
Ten small group discussions were held. For example, one group discussed the prompt, “Assessing data requirements and diversifying data sources for agri-environmental monitoring and management, and overlaps with cultural and natural heritage monitoring and management requirements
- For continuous monitoring data on different soil properties, what is the cadence of updates needed to inform sustainable management?
- What proxies are used in different domains?
- Focus on proxy data provided by sensors. How do they need to be deployed?“
Participants were involved in a wide range of projects. For example, one project monitors illegal tree cutting. Data collected involves temperature, soil moisture, humidity, light (light sensors), noise, and smoke (smoke sensors to detect fire). The challenge for this project is data fusion. Another project monitors the microbiome associated with potatoes and looks at individual soil samples.
In discussing how to collect data together, one of the main challenges identified was language and terminology barriers. Building a common database of terminology alongside the data to communicate across disciplines was suggested. A second challenge identified is how to integrate real-time measurements and physical sensors. Associated with this challenge was the importance of understanding why you’re taking these samples and measurements. This discussion prompted further questions.
Is there an understanding between sectors? For example, farmers may be taking measurements more frequently over a large area while researchers and scientists taking specific one-shot measurements from a specific site or context (differences in frequency and scope). How do we format those differences?
Major questions that could guide integration include:
- What measurements are we taking?
- How are we taking the measurements?
- When are we taking the measurements?
The group also noted the need to decide:
- Where data is stored.
- How to discover and connect groups working on related projects.
The challenges of long-term initiatives were highlighted, noting that we need to think in terms of 10 years, 20 years, or 30 years or more. The group also emphasised that a digital snap-shot is not enough—we need the physical samples as well. Soil data cannot rely on digital sensors alone.
Storage is a significant challenge—for digital data and physical samples Storage can be very expensive, which could inhibit open-source initiatives.
Reporting back
Groups summarised their key points on post-its, which were used in reporting back to the workshop and prompted further discussion.
What makes remote and near surface sensing data more robust?
Robust data is: Well correlated with what you’re trying to measure, collected with sensors that are benchmarked, collected at appropriate intervals.
How to enhance soils data by focusing on biological communities above and below the ground?
Data Needs: We need to collect proxy data to inform on the composition of microbe communities at scale. To do so, we must improve calibration between measurements taken at diverse scales. We also need to improve understanding of sensor placement in relation to the locations of target vegetation and develop protocols for the assessment of change over time.
Barriers: Time and resources.
What variables do we not record?
Data Needs: It’s difficult to measure variables related to biodiversity, soil turnover and soil health at scale. Measurements of colour, compaction, carbon content, organic matter content, ground water are needed, often more than once.
Citizen Science can help to provide some of this data: Soil health kits can be used to collect this data in standard form. These kits could be distributed for use in urban areas, gardens, farms, heritage sites and on rural estates and parks with pathways open to the public.
How do we collect data from remote areas effectively?
Data needs: Samples need to be representative and sampling locations and strategies should be better aligned with communication networks.
Ways forward: When working in remote areas, more care is needed in selecting what information is essential at the point of collection, ideally processing locally and only transmitting what is necessary. Cost effectiveness as a key concern, constraining how many and what kinds of sensors are deployed. For remote deployments, the interplay between hardware, and software platforms is not well standardized.
How can we enable data sharing across domains?
Open questions: Understanding domain and application specific questions better is essential. It’s difficult to understand datasets, particularly large complex datasets, without understanding why they were collected. How can we provide insight into the research questions and context which shape data collection?
How can we distribute data responsibly, widely and directly, while retaining the contextual information which makes it interpretable? What kind of data requires expert interpretation, and should not be distributed ‘raw’?
How do we set up better communication between projects to provide this contextual information?
How do we maintain information about the context and representativity of samples so that we avoid misrepresenting larger areas? How do we ensure this contextual information isn’t so complex and specialised that it reinforces barriers between domains and discourages data reuse?
How do we improve data on soil nutrients and related soil properties?
Ways forward: Increase efficiency of data collection by developing better nutrient sensors to embed in the field, rather relying entirely on lab analysis. Coordinate more closely with farmers on data collection. Work to better collate and integrate existing data currently held by individuals and organisations where possible, rather than collecting new data.
Build better narratives using data on soil nutrients to motivate individuals and groups to invest in collecting and sharing them.
How can we collect data on soils opportunistically and from unconventional sources?
Data collected for monitoring vegetation development and conditions can be used as a proxy for what’s beneath the surface. Care must be taken with calibration over larger areas.
Data on biodiversity similarly can provide proxy information on likely soil types.
Archaeologists could be recording more soil data for other soil scientists during watching briefs, surveys and heritage management.
What data is more challenging to collect?
Soil nutrient analysis and soil structure analysis need to be coordinated. Mapping and monitoring soil nutrients using remote sensing remains challenging, compaction mapping likewise remains a challenge. Access to management practice data is an issue as some growers will not wish to openly share this information.
Closing thoughts
Data is interesting only in terms of how it relates to what we want to find out. The workshop brought together a broad community and the range of interests present and challenges discussed reflects how connected we are to the soil. Whatever we do to meet these challenges, we need to do in a collaborative way.
In building this collaboration, we need to consider what makes land valuable in the broadest sense. We often ask: what is the value of the land? Maybe we should be asking: what should be the value of the land?.
Programme
| Time | Activity | |
| 09:30 | Welcome | |
| Talks (speakers) | Talks (titles) | |
| 10:00 | Rachel Opitz (Archaeology, University of Glasgow) | Enriching Scotland’s soils data – sensing, modelling and interoperability |
| 10:10 | Malcolm Coull (James Hutton Institute) | How do you solve a problem like soil data? |
| 10:30 | John Holland (SRUC) | Establishing a LoRa sensor network at SRUC Kirkton Farm – monitoring the environment and livestock. |
| 10:45 | Elisa Ramil Brick (Heriot-Watt University) | Gaps in research in precision agriculture, precision livestock farming and agroforestry and the need for a new area of research in “Data-Driven Agroforestry”. |
| 11:00 | Marian Scott (Mathematics and Statistics, University of Glasgow) | Data fusion in space and time. |
| 11:15 | Rachael Wakefield (CENSIS) | Wireless coms options for connected devices in remote areas. |
| 11:30 | Ciara Keating (School of Engineering and Institute of Biodiversity, Animal Health & Comparative Medicine, University of Glasgow) Umer Ijaz (School of Engineering, University of Glasgow) and Barbara Mable (Institute of Biodiversity, Animal Health & Comparative Medicine, University of Glasgow) | Impacts of agricultural practices on soil communities associated with potatoes. |
| 11:45 | John Crawford (Adam Smith Business School, University of Glasgow) | The science of delivering soil health at scale. |
| 12:00-13:00 | lunch break | |
| 13:00-13:15 | quick explanation of themes | |
| 13:15-14:00 | thematic group discussions 1 | |
| 14:00-14:15 | swap groups (and top up your tea and coffee) | |
| 14:15-15:00 | thematic group discussions 2 | |
| 15:00-16:15 | thematic groups report back | |
| 16:15-17:00 | closing roundtable discussion |
Participants
| Name | Affiliation |
| Barbara Mable | University of Glasgow |
| Helen Sandison | Censis |
| Umer Ijaz | University of Glasgow |
| Chiara Keating | University of Glasgow |
| Malcolm Coull | James Hutton Institute |
| Marian Scott | University of Glasgow |
| Rachel Opitz | University of Glasgow |
| Ying Zheng | University of Glasgow |
| Daniel Barreto | University of Glasgow |
| Roy Nielson | James Hutton Institute |
| Eamonn Baldwin | University of Glasgow |
| Lisa Brown | Historic Environment Scotland |
| Matt Aitkinhead | James Hutton Institute |
| Mu Niu | University of Glasgow |
| Patricia Bruneau | NatureScot |
| Sarah Poppy | Historic England |
| Vanessa Reid | Durham University |
| Stewart Arbuckle | AgriCarbon |
| Michelle Bloor | University of Glasgow |
| Ashley Stamper | AgriCarbon |
| Claire MIller | University of Glasgow |
| Awais Aziz Shah | University of Glasgow |
| Caroline Gauchotte-Lindsay | University of Glasgow |
| Michael Muir | University of Glasgow |
| Muhammad Ali Jamshed | University of Glasgow |
| Ria Dunkley | University of Glasgow |
| Elisa Ramil Brick | Heriot-Watt University |
| John Crawford | University of Glasgow |
| John Holland | SRUC |
| Rachael Wakefield | Censis |
