Floodplain meadows provide a valuable and rich ecosystem, supporting some of the most diverse vegetation in the UK and performing key ecosystem services such as flood storage and sediment retention – important given climate change. Yet, less than 1500 ha of this unique habitat remain in the UK, which has motivated its protection under the European Habitats Directive (92/43/EEC).
A better understanding of these vulnerable ecosystems will ultimately allow improved environmental management. To this end, the FUSE project aims to investigate the variability over time and distance (sub-metre to tens of metres; half-hourly to seasonally) in the plant community composition and the function of vegetation in these floodplains. This will enable scientists and environment managers to better understand the relationship between this variability in plant life and the key physical variables and nutrient levels of the soil, via an intensity of monitoring not previously attempted in this environment. Such detailed observations are required because plant processes are responsive to fine-scale changes in the environment and short-term seasonal effects. In turn this is important because the phenology of competing species varies; some species rely on high nitrogen mineralisation rates early in the season, whilst others have stored reserves to support spring growth. To be able to understand this complex ecosystem the project is employing sophisticated high-resolution model-data fusion techniques involving a dense array of underground wireless measurement technologies, high-resolution hierarchical observational data, and state-of-the-art modelling tools.
The underground monitoring array consists of battery powered computing devices with sensors, e.g. for soil temperature and moisture, connected via wireless networking. The convergence of miniaturisation of key technologies and the development of low-powered radio communications has enabled the development of a network of autonomous sensing nodes. Designing these Wireless Sensor Networks (WSN) is challenging in terms of being able to guarantee the accuracy of sensor readings, the reliability of data communication, and a guaranteed lifetime that will enable meaningful monitoring periods.
Designing such an array has been a further challenge for the FUSE project because of physical and social constraints. The Oxford floodplain is a Site of Specific Scientific Interest due to its rare mix of plant species, sheep and cattle graze there during autumn/winter, and it is publically accessible. This places limitations as to where the WSN components can be placed; out of sight and out of reach. As a result, the project is required to bury the entire technology (including the antennae) underground thus creating a wireless underground sensor network (WUSN). Further, deploying a low-cost, dense WUSN system in a protected environment means that disturbance to its vegetation must also be minimised. This requires a solution that is low-powered (so that the battery powered nodes can maximise their life-time thereby avoiding the need for regular replacement) with a small form factor, which places limitations on the communications technologies that can be used.
These limitations in deployment present the largest challenge to the project. How can underground to over ground (U2O), and over ground to underground (O2U) communication be achieved between the sensor nodes and the base station for the relay back of environmental driving and verification data to the environmental models? Further, due to limitations on the transmission distance such low-powered transceivers can achieve, a considerable element of underground to underground (U2U) communications is needed to enable messages to multi-hop (relayed by intermediate nodes) across the space.
The design approach chosen came with a large risk. At the beginning of the project the state-of-the-art low-powered WUSNs could only achieve U2U distances of a matter of centimetres. This would have meant placing 1000s of nodes in the ~0.5km2 floodplain meadow, which is neither a scalable nor cost-effective solution. Furthermore, the soil conditions at the sites used for prior research were relatively dry and the soil texture much coarser than that of a floodplain, which can appear like a lake in winter and a lush meadow in summer.
An extensive mix of theoretical calculations and field studies of the effects of dynamic soil conditions on low-powered radio signal transfer was undertaken. This confirmed the extent to which:
Attenuation of the radio (RF) signal is correlated to the spatiotemporal variability in terms of soil composition, water content in the soil, soil temperature, vegetation on the ground etc.
Inhomogeneous soil composition and different vegetation levels on the ground across the field affect channel quality in different regions, introducing the heterogeneity in the network.
Variation in external weather conditions, soil temperature changes and their relationship to time affects channel behaviors.
Performance of other key system components like Clock Oscillator and Battery are affected by varying temperature and humidity levels within the soil.
Exploitation of this knowledge has brought success in that, while remaining in the low-power spectrum the network is able to achieve on average 50m U2U communications distances, with much farther distances between U2O nodes.
With this knowledge, the WUSN topology has been designed using a numerical optimisation procedure based on a geostatistical model, while ensuring both monitoring coverage of the existing subsurface area, minimisation of likely interpolation errors, and taking into account the communications constraints mentioned above.
In addition, intelligent network protocols are being developed to enable adaptive radio messaging that minimises battery consumption while maximising data delivery. By exploiting the soil sensor data, the system ‘understands’ the spatiotemporal changes in the soil and is therefore able to determine its best networking strategy. This adaptability can be in the form of message formats deployed or transmission timing. This approach of understanding the soil medium and how it impacts the network protocol is an example of a cyber-physical system; the project therefore contributes to reasoning about such systems’ behaviours.
WUSNs have the potential to revolutionise our understanding of plant community composition and vegetation functioning in flood plains. Further applications of this research can also benefit monitoring applications for land slippage, pipelines, roads, embankments, train lines, etc. benefitting government agencies, energy companies, water industries, and transportation companies.
The FUSE project is funded as part of the Natural Environment Research Council’s Sensor Network Programme. It is led from the University of Reading in collaboration with Imperial College London, the Open University, University of Southampton, British Geological Survey, Natural England, and the Environment Agency. For further information on the project please contact Professor Anne Verhoef, University of Reading by clicking here.