First of all, WTN’s risk scoring Methodology considers both types of impacts: direct physical damage and business interruptions.

 

Second, we need to consider all bushfire, storm, and flood information within a certain perimeter. The perimeter of interest may be 100 meters or several kilometers and it can differ between hazards, floods vs wildfires, for example. Spatial considerations are also specific to the type of facility, whether we’re talking about railroads, telecommunications, mineral mining sites, or real estate. We always discuss these settings explicitly for each project and use case. 

 

Now that we have all necessary data from multiple observational sources and also from several climate models, we can evaluate the spatial extent of fires and floods and the intensity of each event around each facility. We take all available records to assess the frequency and amplitude of hazards.

 

The core building block of the risk assessment is the classic Generalized Extreme Value model (GEV). By the way, this methodology is also used for insurance and credit risk modeling. The difference is that we don’t apply Monte Carlo (stochastic resampling) to evaluate rare and extreme events: we calculate the GEVs directly from the data. We can do it without Monte Carlo: because we repopulate the raw data sample with abundant climate model time series.

 

Imagine you don’t just use 50 years of observations but also millions of model-derived data points. Then you can calculate the probability of high impact events. 

 

The frequency of extreme events is described by their return period and amplitude. Probability (or likelihood) is the inverse of the return period. Rare, high-impact events have long return periods and low probabilities. 

 

Frequent events (small anomalies) may occur every other year; they have short return periods and high probabilities. For example, if on average a flood event happens once every 10 years, the annual probability of this event is 10%. Figure 1 demonstrates four GEVs for 4 risk levels: moderate, high, severe and extreme. You can see that it is much like the classic Risk Matrix: with intensity on Y-axis and probability on the second axis. 

 

Each curve on Figure 1 represents the function that describes the behavior of rainfall anomalies at some location. It begins with frequent, smaller anomalies that have low impact but high probability (left-hand part of the curve). These curves tell us, for example, that once every 25 years rainfall may reach 150 - 300 mm/day, depending on the location, and once every 100 years it may reach 200 - 500 mm/day, and once every 200 years it may reach 600 mm/day for one of these locations. This kind of information is critical for urban drainage system design, flood defense construction, and hydropower security planning. 

 

High-risk events are exceptional. They happen rarely, which means they have a low likelihood of occurring, but when they do, they cause major impact. In contrast, events that occur frequently (every day or every year) have a high likelihood but low impact. These are low-risk events.

 

 Important to note here : low probability does not mean low risk!

 

 

Figure 1. Example of GEV curves for four locations with different risk levels: moderate, high, severe, and extreme. Y-axis: normalized anomaly (amplitude). X-axis: annual probability.

Locations with low risk are outside the scope of GEV: there are no extreme values.