Our selected case studies show the expertise of our team and examples of projects we have and can support you with.
New technologies for monitoring livestock behaviour
This project is a collaboration between Aberystwyth University, the RSPB and the Elan Valley Trust. A combination of new and existing monitoring methods using fixed-wing Unmanned Aerial Vehicles (UAV’s; drones) as sensor platforms are being developed to track the movements of livestock animals on large-scale (~100ha) hill enclosures, quantify and map the vegetation present, and identify key features which might influence animal behaviour (e.g. water bodies, pathways etc). The understanding of how these different species interact with the landscape around them will then be used to develop interventions which target their grazing on specific areas (e.g. Molinia/ Purple moorgrass on peatlands).
A key objective of this project is to utilise low cost equipment which is easy to use, so these techniques may be utilised by others. This has included the deployment of fully customised GPS loggers, and the validation of the first custom built system of animal tracking that uses radio-frequency identification (RFID) tags and a UAV receiver.
These tags, which are placed on either the horns or ears of the animals are a low cost (~£11), low maintenance, alternative to GPS collars that work by transmitting individually recognisable radio signals to a UAV receiver, which triangulates each individual tag’s position using its on-board GPS. If proven successful, the flight range of UAVs together with the low cost of the tags could allow large numbers of livestock (100s) to be monitored on over large (150-200 ha) areas.
Modelling Site Suitability of an Invasive Species - Impatiens glandulifera
The risk of invasion from alien species is increasing with globalisation and climate change. The main impacts from plant invasions include alterations to soil chemistry, hydrology and fire regime, biodiversity loss, costly management schemes and infrastructure damage.
Therefore, effective eradication schemes for invasive species are required. Despite this, determining locations to target schemes is challenged by a lack of knowledge in the species establishment extent.
To address this an online database and Maximum Entropy modelling technique was used to derive potential species establishment within Wales. Occurrence data from the National Biodiversity Network Atlas and environmental data were inputted within a species-tuned Maximum Entropy model.
The model highlighted flood areas to be the most suitable growing environment for Impatiens glandulifera whilst areas within the Cambrian Mountains, exceeding elevations of 300m, contain sites with the least suitability.
Recommended eradication method is for volunteer groups to manually hand pull, hoe, or strim catchment areas starting from the source of watercourses highlighted as high priority in the model. This should be annually, preferably after flood events and before plants develop seed in August.
Using Spectral Heterogeneity to predict Species Diversity and Historic Fertiliser Application - Dŵr Cymru Welsh Water
Welsh Water wish to address the issue of diffuse pollution from agricultural fertilisers, with a company objective of increasing the water ingress quality in supply areas. To monitor and collect soil data on an area this size via field excursions would be impractical. Therefore, a method including the use of drone imagery and the creation of a methodology which utilises plant biodiversity as an intermediate latent variable to provide spatial data on nutrient (especially Nitrate) accumulation in farmed pastures are proposed. In this feasibility study Treberfydd Farms and, Newton Farm, Wales, was selected as the study site, with the intention of providing a proof of concept for the use of spectral heterogeneity as a proxy for the detection of excess nutrients on farmed grassland.
Outputs from the study revealed congruence between pixel heterogeneity recovered from remote sensing imagery and plant species biodiversity recorded in the field. The work therefore demonstrates the potential feasibility of using pixel heterogeneity methods to monitor the spatial location of nutrient over-application or accumulation (via run-off/leaching) within a farm environment. Future projects should seek to calibrate spatial resolution of the approach and to characterise the relationship over a wider range of landscape terrains and management regimes.
“The GEOM Team at Aberystwyth University provided Welsh Water and its partners with a methodology to calculate species diversity within a field from drone imagery. The overarching theory is that an increased application of fertilisers is associated with a decline in species diversity. Thanks to the thorough research carried out by the GEOM Team, Welsh Water has the tools to look at taking this one step further to identify intensively farmed areas at a catchment scale and work with farmers to tailor the application of nutrients to minimise diffuse pollution.” – Shaun Lewis, Spatial Risk Analyst – Catchment Team for Dŵr Cymru Welsh Water.
Species Distribution Modelling of the Red Squirrel in Wales - The Wildlife Trust of South and West Wales
The Wildlife Trust of South and West Wales works with volunteers and communities to protect habitats and species, with an objective of building an environment rich in wildlife for everyone by rebuilding biodiversity and engaging people with this environment.
A feasibility study with the initial aim of providing context and producing a workable methodology for species distribution models of Red Squirrels was undertaken. A multi-model approach was implemented, whereby a comparison could be made of overlapping habitat deemed suitable by each model to provide further reliability of suitable habitats for the Red Squirrel.
Results generally indicated the variables to have a weak response on Red Squirrel distribution. Of the variable deemed most important (Grey Squirrel proximity), there is a concern that bias within Grey Squirrel sightings may well be the cause of the observed response (that Red Squirrels are found more often when more Grey Squirrels are present), and as such the results not valid. Models were also run with proximity to Grey Squirrels excluded as a variable, however the effect on the importance of other variables were negligible. It is likely that meaningful results from the species distribution modelling maybe achieved if systematic sampling methods are used for sightings data collection for both red and Grey Squirrel were made.
A processing environment and framework has been generated to reduce the expertise required to initialise the modelling procedure. As such repeat iterations of the modelling process may be undertaken by Wildlife Trust of South and West Wales (WTSWW) as new sightings data emerges, which over time should improve the accuracy and validity of the modelling results. Outputs can then be further verified through the expertise of the Squirrel Partnership and the analysis of the contributing variables. Overall, this will provide the WTSWW with an insight into which environmental variables contribute the most to suitable habitat for Red Squirrels, and where this habitat is in order to focus conservation strategies.
“Working with GEOM helped us make the most of what would otherwise have been lost time due to COVID in 2020. The project enabled us to find gaps in our datasets and focus the project going forwards” Sarah Purdon, Mid Wales Red Squirrel Officer of The Wildlife Trust of South and West Wales.
Data Collection Viability Assessment - Wales Coastal Monitoring Centre
Drone flight viability in coastal and estuarine regions is dependent on favourable metrological/tidal conditions and model/sensor-specific limitations. GEOM (in partnership with Wales Coastal Monitoring Centre – WCMC) developed an automated technique for determining fieldwork suitability (ground and aerial) across coastal regions in Python. The purpose was to improve continuous monitoring of the coastal zone by: 1) Determining typical tidal-meteorological conditions of a given coastal location through time-series analysis; 2) Supporting decision-making regarding investment in new survey equipment; and 3) Minimising risk in the use of such equipment.
The technique involves combining ERA5 reanalysis data (ESA) with tidal readings from the British Oceanographic Data Centre (BODC) within a database at the hour scale. This allows the user to estimate the total/average number of viable hours per month for a given piece of field equipment (drone or ground survey). Database searches are parameterised by the user to allow viable hours to reflect equipment capabilities (e.g. maximum wind resistance) and the optimal astronomical/meteorological conditions required for data collection (e.g. exposure of the inter-tidal zone). The use of ERA5 reanalysis data as a global gridded dataset allows the procedure to be applied across any coastal location, depending on the availability of concurrent tide data.
“GEOM has a highly motivated and technical team who clearly understand our unique challenges and through close communication were able to deliver a software solution to support our needs.” – Gwyn Nelson, Programme Manager at Wales Coastal Monitoring Centre.
Priority Habitat Mapping - National Trust
The National Trust UK is Europe’s largest conservation charity and cares for:
- Over 780 miles of coastline
- More than 250,000 hectares (ha) of land
- Over 500 historic houses, castles, parks and gardens
- Nearly one million works of art
As the UK’s largest private landowner, addressing the decline in British animal species has become a top priority for the organisation. The UK Biodiversity Action Plan lists those habitats which were identified as being the most threatened and requiring conservation. National Trust have set an objective to create 25,000 ha of new priority habitats by 2025. This means 10 percent of the land owned by the National Trust will be dedicated to restoring priority habitats and 50 percent of their farms will become nature-friendly 2025.
Adopting practical approaches, such as field excursions, to identify suitable land on this scale would be impractical. The National Trust’s in-house GIS specialist had identified the potential of using Earth Observation however were unsure of the best ways of implementing the data collection nor analysing the colume of data that would be gathered.
The National Trust and the GEOM project team developed the project further and proposed a method of using satellite imagery and framework which would automate a number of the steps of identifying priority habitats. The projects intentions were to:
- Develop a framework which utilises satellite imagery, ancillary data and training data to create a machine learning model used to classify habitats
- Create a priority habitat classification map of the National Trust real estate
- Identify 4,600 ha of priority habitat so that the National Trust can include it within their designated protection zones
- Identify area of non-priority habitat so that the National Trust can implement habitat restoration plans.
The project was successful in developing a framework for identifying priority habitats to be included in The Trust’s designated protection zones. A total of 6,693 ha of potential land was identified across Wales that could be used to create or restore priority habitats and demonstrated that this approach can accurately identify 28 of the 40 terrestrial priority habitats. Validation of the results showed that the map which was produced had an overall accuracy of 82.46 percent. The study also identified how future improvements could be made to the method by incorporating further data sources.
The potential impacts of the project for the National Trust are wide ranging. It provides evidence that supports the accuracy of adopting such an approach that will enable more efficient identification of priority habitats. If this is scaled up across all regions of the UK, it may prove to make a valuable contribution towards The Trust meeting its objective of identifying and creating 25,000 ha of priority habitat in the UK.
A low-cost data transfer system to facilitate sensors on livestock farms
To this date, considerable academic research has gone into the development of sensors for use on sheep and cattle. However, with regard to their application on farms they have been limited. Sensor cost, technically difficult user interfaces, and issues with data transfer from remote locations on the farm to a centralised server have made many sensors unappealing for adoption into general farming practices.
In collaboration with Caerphilly council, IBERS and the Computer Science Department at Aberystwyth University are in the process of creating a data transfer system which can successfully relay data from animal, and fixed based sensors from field to a central server stored on farm. The objective being to produce a working example of this system complete with sensors, receivers and integration mounts. Emphasis is placed on deriving as lower cost components as possible for the hardware, and that the system be autonomous in nature.
In doing this, we can create a universal platform to which different types of senses can operate on seamlessly, and thereby allow flexibility in the choice of the sensors that individual farmers wish to adopt without needing to alter the main infrastructure. Furthermore, by developing such a system, we can reduce some of the functional requirements of the sensors (e.g. omitting the need for large stable storage capacity on each unit), thereby lowering the cost of the sensors.
A Tourist Guide to Nambour’s Past and Present
As part of sharing the rural responses to globalisation, Global-Rural project designed and created a web based “walking” guide re-enacting the making and remaking Nambour, in rural Australia through raw cane sugar.
The content based on archival research and primary field work was translated to provide users with linear, place-based narrative featuring multimedia – videos, audios, interactive and static maps, images, photos etc. Each geo located place is accompanied by a narrative highlighting events about the location on the town’s struggles through the cane sugar industry. The narrative guide also includes places of interest, heritage sites and attractions.
For example, locations and videos of the last sugar crush, last burn down of cane plantation at Moreton mill, audio interviews with residents and former workers, videography etc. The guide can be followed sequentially through each story point, browsed interactively through the map or through the thumbnail carousel. For example, the survival of Nambour after the closures through conversion of its sugar trams, rail lines, plantations, staff accommodations to real estates, shopping malls, tourist attractions can be searched with key words.
The platform is created to serve as a tour guide while in Nambour. It is also a remotely accessible virtual tour based on a web browser and designed to be downloadable as a print or work on mobile devices. Features can also be reconstructed as 3D experiences.
Object Detection of Stones Within a Stone Wall - Royal Commission on the Ancient and Historical Monuments of Wales & Dyfed Archaeology Trust
The Royal Commission on the Ancient and Historical Monuments of Wales (RCAHMW) and Dyfed Archaeology both work across Wales alongside professionals and volunteers to help protect, record and interpret aspects of the historical environment. This project focuses on aiding the efficiency of completing building surveys, specifically to minimise the human input needed to extract and count the number of stone objects within a stone wall.
The initial project aim was to produce a workable methodology contained within a user-friendly software to detect and extract stone objects from an image of a stone wall. Through reviewing a variety of differing software and methods, a combination of analysis procedures where utilised and contained within a single executable file.
This single executable file has been generated to run on windows machines without the need for installation of the application itself or any dependencies. The application loads a Graphic User Interface when opened and runs through a guided workflow to enable users to extract stone objects from loaded images of stone walls in the format of an AutoCAD DXF format file.
“The development undertaken so far in the vectorisation of individual stones from an image of a monument and then output them an AutoCAD drawing has huge potential in speeding up what has traditionally been a very time-consuming process.” – Scott Lloyd, Fundraising and Project Development Coordinator of the Royal Commission.
“The application developed as part of the project has the potential to rapidly produce stone-by-stone drawings and so speed up what is traditionally a slow and tedious process.” – Ken Murphy, Chief Executive Officer of the Dyfed Archaeology Trust.
Baseline Mapping Module - Sazani Associates
Sazani Associates are an international not-for-profit research and development organisation. They have teams based in Wales and Zanzibar and work in various fields such as socio-economic development, natural resource management and education.
Sazani are working with existing client, New Liberty Gold Mines (NLGM), to monitor changes to land use/land cover (LULC) as a result of mining activities in Liberia. They seek to carry out historic mapping and continuous monitoring of LULC to understand the changes and impact to biodiversity as part of the Bae Mountain Development Agreement Property Licence. This area lies within the Guinean Forests of West Africa; a biodiversity hotspot which is home to 9000 plant species, of which 1800 are endemic to the region.
Biodiversity offsetting is a compensatory approach to mitigating the environmental impact of developments and is well suited to large-scale, open pit mining operations such as those in Liberia. This type of mining is associated with habitat loss, disruption of aquatic habitats, and the subsequent bioaccumulation of toxic chemicals within the food chain. Developing countries are particularly vulnerable to these impacts due to the economic potential of the mining opportunities. With an increase in mining activities within biodiversity hotspots, biodiversity offsetting is important for ensuring there is ‘no net loss’ as a result. Establishing a biodiversity baseline is a crucial step in the development of a biodiversity offset plan and provides the means to measure the impact and potential loss of biodiversity as a result of the mining activities.
The collaborative project between GEOM and Sazani Associates aimed to develop a novel framework for the use of automated Earth Observation (EO) techniques to assess habitats loss close to mining areas. These techniques built into a Baseline Mapping Module (BMM) in Google Earth Engine platform, uses historical archival earth observation data in the platform for the analysis and monitoring of vegetation cover dynamics associated with mining. The framework is designed to be user friendly, with minimal need for user input and reduced data processing time while deriving baseline information of the study area.
Once developed, the BMM was trialled in the NLGM operation by investigating three land cover metrics: Percentage land cover distribution; vegetation loss; and the spatio-temporality of vegetation cover. The tool clearly identified the impact of the NLGM operation on the surrounding environment by estimating vegetation loss of approximately 2.22 square kilometres. The project demonstrated the feasibility of an automated approach in deriving land cover metrics.
The BMM has the capacity to incorporate new images as soon as they become available on Google earth engine platform and this has made possible the continuous monitoring of ecological baselines using historic data.
“The land use change app developed in partnership with GEOM was a proof of concept product enabling the mapping of land change over time using remote sensing data. Once the concept was engineered in partnership with GEOM, Sazani was able to take the learning and moved to reframe the application through the investment of ESRC funding with Swansea University and a range of external developers. Without the time spent working with the GEOM team this platform work not have emerged. Sazani are using a land change mapping tool developed from the learning with GEOM to identify how to plan a biodiversity offset in Ethiopia, supporting communities to develop climate state livelihoods in an area undergoing major resettlement very close to the South Sudan border.” – Mark Proctor, Director at Sazani Associates.
Participatory Mapping Mobile App Development and Design: A Modular ODK Framework - Sazani Associates
Sazani and GEOM collaborated to explore options to develop a mobile participatory mapping app. Together, they drew up a list of proposed capabilities the app should have that goes beyond a simple digital survey to include features such as mapping, tracking, geotagged photographs, and multiple-type questions and responses. The mobile app aimed to provide Sazani with the means to gather person centred data on participants demographic characteristics, and their environment in a way that is confidential, interactive, confidential and user friendly.
The GEOM project team carried out research into the platforms that would support data collection by surveys with mapping-led questions whilst Sazani carried out research into citizen science-led features. Sazani also developed a set of citizen science-led questions which would form the basis of designing and configuring a mobile data collection app. It was important that all parties worked to ensure the app was developed in line with best practice in participatory planning and within Welsh planning law. The project adopted an ODK framework.
The outputs from the project include the Sazani Participatory Mapping App operational on the android platform; the web form version of the mapping tool; the authored Sazani Participatory Mapping XForm containing the questions and features requested; and the free and open-source code for the app.
The development of the Sazani Planning Mapping App has the potential to revolutionise the way planning data is collected to contribute towards participatory planning. The features of the app and web version make participatory planning more accessible, and cheaper than standard methods and, as a result, it is hoped that greater community engagement, particularly from marginalised groups, can be achieved in future planning projects.
The creation of the app enables Sazani to offer a new service in any future citizen science community-based planning projects. The intention is that, as the app grows and its use is adopted more widely, this will create opportunities for Sazani to employ more staff.
Companies We’ve Worked With:
What We Do
GEOM’s goal is to support companies and organisations with state-of-the-art spatial intelligence towards developing market-ready products and services
10 to 250 employees
Turnover less €50M
SMEs in West Wales and Valleys
Any Partner Organisation
Environment and Conservation
Tourism and Cultural Heritage
Utilities and Infrastructure
Transport and Logistics
Type of Support
Spatial Data Analytics