An invitation: help HEPEX program apps to teach the value of probabilistic forecasts

Contributed by Florian Pappenberger, Andy Wood, Maria-Helena Ramos, Schalk-Jan van Andel, Louise Crochemore, and Louise Arnal

This is a blog for people who usually do not read this blog. This is a blog which asks for help so that our learning and teaching games and tools become more useful and widespread. This is a blog hoping that we can find some enthusiastic volunteers.

The more people know how to use probabilistic forecasts, the better their decisions and the lower the impact caused by extreme weather or hydrology (floods, droughts etc) will be.

Weather, water and fire (and many others) forecasts are uncertain — people instinctively know this. For example, if I forecast a maximum of 15 degrees Celsius for tomorrow, you probably know it will not be exactly 15 degrees Celsius but something around that number (and you may add an uncertainty range from your experience).

Most weather and water forecast centres also understand this and produce not just a single forecast but a set of forecasts. For instance, the Global Flood Awareness System generates 51 scenarios (51 different forecasts) of what the future could look like. This enables it to issue a probabilistic forecast, which means (for the above example) that you can estimate a certain percentage chance of the temperature being above or below 15 degrees Celsius.

Despite this uncertainty, people still need to take decisions. An example of a simple decision is whether or not to take an umbrella when you leave the house. A more complex one may be whether you should stop the London Underground because of a risk of flooding along the Thames.

Probabilistic forecasts enable better decisions because you know what the uncertainties are, whereas you have no clue of them with a single forecast.

Forecast users can strengthen their understanding of the value of probabilistic forecasts, and their skills in using them, through forecasting games. There are many examples:

• controlling a reservoir on a river to avoid droughts (here),
• exploring the economic value of uncertain forecasts (here),
• managing floods (here),
• weather roulette (here) which evaluates the information in forecasts.

HEPEX has designed and implemented a number of such games (see here), but, to date, except for our first try on the online version of the “Pay for a forecast” game (here), they are paper- or presentation- based, and we would like them to reach a wider audience.

The challenge today is:

1. Can you improve the existing ideas?
2. Could you design a game where different players compete against each other? (*)
3. Most importantly, can you transform the paper-based version into apps (web or phone) to reach a wider audience?

We are also interested in putting together ideas on how to set up a project (volunteer, student, commercial), which kind of partners, and how to get funding (volunteer-based, crowd-funding, H2020 and the like, commercially, etc) to enhance our tools for training and teaching the value of probabilistic forecasts.

Please contact us or any of the HEPEX co-chairs by Friday 28th April if you could contribute to engineering web-apps in support of the HEPEX initiative or of you are interested in participating to project proposals on the topic.

You can also come and meet us at the EGU poster session of the ensemble session in Vienna. Our HEPEX poster, will be displayed on Friday, 28th April, and the attendance time is from 17:30–19:00 in Hall A.

(*) In the Weather Roulette game, for instance, you would need to decide whether the app decides the odds for different outcomes. The goal of each player is to win the most in the weather roulette casino shared by everyone. Alternatively, players could each have their own casino and bet in the casinos of other players. Examples of the questions a designer would face include: On what forecasts should the odds be based? And what information is made available to the players so that they can decide how much to bet on different outcomes?