When do I need to switch from a single design storm to an ensemble of storms?

You might say “I’ve been doing drainage design with rational method, or an SCS design storm for a very long time, why do I need to change?”

That depends on the type of problems you are trying to solve. As many places are getting more urbanized, more and more developed communities are having more serious flooding issues. And when using the old design methods to size a solution, it usually will require a lot of land that is no longer available and a cost that is simply too high.

The tough reality is that we know we cannot completely fix flooding issue in developed communities in most cases. We can no longer say we’ll build whatever it takes to fix the issue. All we can say is that we’ll try our best to find an affordable solution that can greatly reduce the flooding risks.

And this requires us to answer a very different question to make the decision. I don’t want to spend too little, then I’ll have flooding issues every other time it rains. I don’t want to spend too much either, then all my money is wasted if the big storm I designed for never comes.

I found it sounds very familiar. This is the kind of questions I ask myself when buying a car insurance or a home insurance. I need to know exactly how much insurance my premium can buy me.

Ensemble storms can help us answer these questions.

Estimate the risks

Let’s take a step back and think through the problem.

• say if the system we plan to build will last 50 years, we need to select from a few options ranging from $10 million to $30 million, to make a decision we need to calculate how much damage each option can help prevent from flooding.
• if I have a crystal ball to tell me what will happen in the next 50 yrs, this is a simple problem to solve. For each option, I can simply play the movie and tabulate my spreadsheet with flooding damage. By comparing with the option of doing nothing, I will know exactly how much damage my investment can save me for each option.
• unfortunately, we don’t have a crystal ball. In this case, how are we going to make this decision? when working with lots of uncertainties, a better question to ask is which of the unknowns if not carefully considered, might lead me to choose the wrong option? This can help us to quickly narrow it down to the important factors to consider.
• an obvious uncertainty is the temporal distribution of the storm. We all know no two storms are the same. Simply because my design passed the test of the 2yr design storm, doesn’t mean I am protected for any 2yr storm in the real world. A true test will be to run the model through 50 yrs or even 100 yrs of actual rainfall data, and then use the results to calculate the actual level of protection I have.

The ensemble storms give us a tool to quantify risks using real world data. Statistical ensemble helps the modeler get a feel of the possible output of the model as the rainfall changes, thus helping the modeler to be more certain about the uncertainty of level of protection the design can provide in the real world.

NOTE: For a more robust analysis, we can use the Monte Carlo method. For example, we can have a simulation of 1000 yrs, by randomly select the modeling parameters based on our best knowledge of their value distribution, we can derive the most likely outcome, and quantify our confidence. For example, we can say we are 90% confident, the savings will be between $15–20 million. If we are less certain about the modeling parameters, the range might be $10–30 million.