Blog credit: Louise Darroch

The Internet of Things, or IoT, refers to the billions of physical devices around the world that are now connected to the internet, all collecting and sharing data. Thanks to a handy cellular modem, an evolving data broker system and the cloud, our Coastal REsistance: Alerts and Monitoring Technologies (CreamT) project has done just that with coastal hazard flood sensors! We’ve connected our otherwise “dumb”, wave overtopping sensors to the internet, enabling them to share their data in near real-time to public desktop and mobile devices without human intervention.

How does it work?

It’s pretty simple really (Figure 1). We’ve adapted the National Oceanography Centre’s novel WireWall sensor (https://noc.ac.uk/projects/wirewall) which is used to detect coastal wave overtopping. The WireWall is 3.5m tall and uses capacitance wires which sample at a frequency (400Hz) high enough to capture the transient events (Figure 2). 

By integrating an on-board computer to do some preliminary data processing and a cellular modem, the sensor can connect to the internet and publish its own data. The sensor essentially drops its data directly into to an Application Programming Interface (API) hosted at the British Oceanographic Data Centre (BODC) (https://linkedsystems.uk/erddap). The communication uses secure standard internet protocols (https) helping to maintain the integrity of the data transmission. The API is NOAA’s ERDDAP and takes care of all the indexing internally as well as transforming the data to a variety of formats on demand including CSV, NetCDF and JSON (https://linkedsystems.uk/erddap/tabledap/CreamT_747f_cj65_2fgh.html). It also enables us to enrich it with Findable Accessible Interoperable and Reusable (FAIR) data principles (Wilkinson et al. 2016), such as controlled vocabularies. Our API is publicly accessible and uses Restful communication protocols. This means our scientists can directly pull the data over the internet into their own code to check on the performance of the sensors in the ways they want. The overall latency from sensor measurement to data access is approximately 10 minutes. We have successfully demonstrated this system with deployments of WireWall sensors at two high-energy wave overtopping coastal sites (Penzance and Dawlish) from March 2020 to March 2022.

To help put our overtopping data into context, we also pass near real-time environmental data (e.g. wind, waves, sea level) from nearby sensors into our ERDDAP API (Figure 3). This is thanks to external APIs provided by the Environment Agency, Met Office and Channel Coastal Observatory.

The overall effect is that it is now possible to plug ‘n play WireWall sensors to detect wave overtopping where strategically needed. For example, at locations that can help inform the design of sea defences or locations that validate flood forecast models (e.g. University of Plymouth SWEEP OWWL model). Most importantly, its nowcast hazard warning capability could be used to inform coastal communities and transport operators of any immediate risks of wave overtopping.

A public viewing…

APIs are nice because they abstract users from the complexity of the data. However, my Mum couldn’t use one, let alone understand what any WireWall data means should she get some. I could teach her but it would take a lot of patience (and gin!). To demonstrate how we can display the data in a way that can help make people like my Mum aware of hazardous wave overtopping, we developed a web dashboard that can be viewed on common internet browsers (https://noc-coastal-hazards-explorer.app).  The CreamT Coastal Hazards Explorer dashboard uses a simple traffic light system to relate the potential risks of overtopping to the public using the data from the API (Figure 4).

We are also developing a prototype mobile app. to reach more people on more platforms. The soon-to-be-live Trueflood^uk app. is aimed at providing real-time multi-flood hazard information to the public (Figure 5, Figure 6). It enables users to view the current UK environmental observations close to a selected reference point. It uses the latest flood related information from external data sources (Met Office and Environment Agency) and displays them alongside our wave-overtopping sensors and 24 hours of the National Oceanography Centre’s anyTide tidal prediction model.

The prototype is designed to be mobile responsive and will be hosted on Google Cloud where it can be accessed live and tested on desktop and mobile internet browsers. Once fully operational, we aim to wrap our app. in some mobile-friendly code developed by Facebook, which we can then deploy to the various app stores. Our long-term aim will be to enhance the app. with real-time flood warnings as well as citizen science supplied ones.

Who needs power?

A drawback of our ERDDAP API is that it only accepts sensor data transmissions in ASCII format, while our WireWall sensor works in binary.  The conversion of data from binary to ASCII can be quite power hungry at the sensor meaning we need to change battery packs every 3 months. To help conserve power at the sensor, we are partnering ERDDAP with an auxiliary Flask API (osd2ERDDAP) which does the binary to ASCII conversion at BODC before posting the data to ERDDAP (Figure 7, Figure 8).

The sensor still uses secure standard internet protocols (https) to publish its data. The only change is the internet address it sends it to. The auxiliary API is currently under development and we have successfully demonstrated communication between a test sensor using binary transmissions and ERDDAP.

Let’s take control!

Since we’ve connected our sensors to the internet, couldn’t we also use the internet to control them? In principle we can! We’ve created a dataset of operational triggers in ERDDAP which we hope will be monitored by our WireWall sensors (https://linkedsystems.uk/erddap/tabledap/CreamT_NRT_Triggers_16c9_cbb4_e94b.html) in the future. Not only can we manually issue triggers to direct a sensor to turn on or off, we’re also attempting to automatically predict overtopping events to help operate our sensors without human control! We are currently developing an algorithm that assesses the change in data from nearby environmental sensors (Figure 9) and automatically issues ‘on’ triggers when it thinks the conditions may support wave overtopping. Trials are also underway to check that a sensor can use binary transmissions to read the triggers from ERDDAP using our osd2ERDDAP API. Being able to remotely operate WireWall will also help us to conserve its power and reduce data storage by minimizing monitoring of non-overtopping events.

And Finally

If you want to find out more about our development with WireWall then please follow us on twitter @WireWall_NOC.

Acknowledgments

The work was funded by NERC through: the Digital Environment programme (CreamT project NE/V002538/1) and support from the ECO MAD (Enhancing Climate Observations, Models and Data) tasks focusing on ‘Digital Solutions for sensor communications’ and the ‘TrueFlood App development’.

References

Wilkinson et al. 2016. https://doi.org/10.1038/sdata.2016.18