Using twitter to get ground truth on floods: an interview with Floodtags founder Jurjen Wagemaker

6 min read
You are a water engineer by background. How did you come up with the idea for Floodtags?
As a water engineer, it has always been a great challenge to understand the situation on the ground (“the ground-truth”) in a very limited time. I experienced this first hand in my first flood project in New Orleans in 2005: local observations and photos from citizens proved essential in reconstructing the flood event. We needed this to understand what had failed in the first place and then how the flood spread. This pattern kept repeating itself over the next ten years where I worked on a number of flood management projects. In one recent project, the Jakarta floods of 2013, the best piece of evidence came from an operator shooting a picture with his blackberry at a critical point (see figure 1)! That was the missing piece in the puzzle. And I can think of many more examples. 

Figure 1 – 17/01/2013 08.00 AM: The Sunter Polder is taking in water from the Kali Sentiong (by operator)

Until recently, the downside has always been that getting these observations has been a slow, laborious and biased process. With the arrival of the social media, getting local content has all of a sudden become easier. And that's where Floodtags come in.

What kind of insights can Floodtags provide to disaster response agencies and development organisations that they would not be able to get through more “traditional” means? Can you give us an example?
Traditional means such as monitoring and modelling are tremendously important. They are the basis of any flood information system. However these means have also their limitations, as they don’t provide a direct window to what is happening on-the-ground. With Twitter, people are actually providing that window and their observations are coming in 10 tweets per second! 

Figure 2 – People are often very specific in their tweets, like where the floods are and how deep it is (on 29/01/2014)
By all means, Twtitter has proved to be a fantastic flood monitor tool and we encourage people to share even more of their flood experiences on Twitter. Now the difficult part is, to create the right flood filters and enrichments, so that disaster managers only need to look at a fraction of the hundreds of thousands of observations coming in.
So we enrich and analyse all flood data in real-time, and present them in an understandable format through our web service. A good example is the water depth of a flood. It turns out that a large number of people both mention the flood depth as well as the location where they monitored it. Take for instance January 29th, 2014: out of the 360.000 tweets we collected on floods, 15.000 included water depth observations (see picture). Together with the Dutch water management institute Deltares (@arnejanvl) we are working to develop a sound interpretation framework for these observations to create real-time floodmaps. For reference, to make a reliable floodmap of the January 2013 flood took a total of nine days. This was thanks to the hard work of the disaster management office and the HOT team (Humanitarian OpenStreetMap Team).

Figure 3 – The approximate locations of 15.000 unique water level observations on 29/01/2014 (amounts of tweets in the circles)
Another example of getting the best from the data is our work together with SEPAM (a.o. @mwieriks), where we combine social network analysis and data analytics to identify “Flood Twitter Champions”. With this, flood and disaster managers can effectively engage with the public and target interested individuals. Better community connection for better community resilience. 
You put quite a lot of thought into your business model. Perhaps it can be described as “freemium” – free basic tools for government but pay-on-demand for added value data analytics. Would you agree? If so, how do you think this guarantees sustainability? 
Our starting point was that we weren't going deep into developing technical solution until we had a sound business model to sustain it. I've seen too many great studies ending up on the shelf, pilots that run out of money and just evaporated! Our firm belief at Floodtags is: whatever problem we are going to technically solve, it should stick and never let go!
Our model is that we give away real-time data for free, so that affected people and supporting organisations always have access to life-saving information on floods. This includes flood maps with water depths, flood alerts and overviews of flood impact and humanitarian response. Everything you need for an efficient handling of water and disaster management operations. And we don't limit ourselves to Indonesia, but we offer this as a worldwide web service, now in 21 languages. Where we earn our money is from advance analytics of the data for our costumers. This can be deeper data mining, enrichments with other data sources, research to historical flood events and so on. And when we work on a project with a customer, we ask whether the scripts developed for them can be incorporated in the free real-time data service! So the circle is complete and we can potentially continuously provide new features and information as a public good. 

Figure 4 – Screenshot of Floodtags explorer
In your experience, what are the greatest challenges (and opportunities) for the public sector in adopting tools like Floodtags?
One of the challenges that the public sector in the Netherlands also faces is to restrain the tendency to develop and host IT tools like Floodtags on-site. Such big internal projects are difficult to manage and usually lead to high costs (that cannot be spread over more customers). This is the reason that we set-up Floodtags the way we did and this is also the big opportunity for the public sector: by using web services like ours, the public sector can carry out their core tasks more effectively, at less cost! Of course there are many people in the public sector who understand this very well, and these are exactly the people that we are talking to right now. 

Figure 5 – A large number of assistance tweets from government, NGOs and various sorts of community initiatives during and after flood events
What is the current status of Can we already use it – and what partners are you looking for?
As per November 27 you can! We will launch the site at the upcoming Data Innovation for Policy Makers conference in Bali. And from that date onwards you can use Floodtags to get realtime flood information in Indonesia. Just go to and sign-up. Especially when it rains it can become quite interesting: you can search for different neighbourhoods and see what people tweeted and how deep the water is. There is also a realtime tweet density map and you can request tweet statistics (e.g. figure 5, where we compare flood tweets with flood response tweets) – and we have got so much more to come. For those who want to connect to Floodtags professionally:
  • For governments and NGOs in Indonesia: You can already sign-up and monitor the floods by our search query and filters. If you want more, such as a configuration of the map for your region, just drop me an email and we will connect you!
  • For universities and knowledge institutes with research in this field: Please email us to see if we can cooperate to improve the service delivery to flood and disaster managers in Indonesia and elsewhere in the world (and meanwhile increase the visibility of your research for the public good).
  • For developers: If you already have a great website and hesitate using Floodtags for its front-end, please note that you can connect to our API and use our data and functionality in your website! As an example, we connected the Floodtags API to J-FEWS (Jakarta’s flood early warning system under the Ministry of Public Works) so they can monitor Floodtags via their existing information system.
  • If you are a citizen: Please sign-up and use the service to help your fellow citizen cope with the floods. You can do so either individually or by teaming up with a local organisation in your neighbourhood. Together we can be better prepared for floods! 
Jurjen Wagemaker of Floodtags can be reached through or @jurjenwagemaker or via the company Twitter account @floodtags.


Did you enjoy this blog post? Share it with your networks!

News, thoughts and ideas about big data and AI, data privacy and ethics from across the Pulse Lab Network. Read more on the blog.

Pulse Lab Kampala

Dialogue is the Key- Shaping AI for Africa

At UNGP, we believe in dialogue. This is why we participated in the Conference on the State of Artificial Intelligence in Africa (COSAA) held in March 2023. The conference was

Scroll to Top