PCRIC’s tropical cyclone product was redesigned in 2024 in collaboration with the risk modelling firm Reask and the global (re)insurance broker WTW, with financial and technical support from the World Bank Group. PCRIC’s approach uses two separate triggers, with the final payout being the maximum payout calculated under the two triggers. Reask, as a specialized modelling firm for tropical cyclones, serves as the calculation agent responsible for determining if a payout is due following a tropical cyclone event.
The SiS trigger is based on the track and intensity of a cyclone and is designed as follows:
- Defining the Insured Zones: Each country or insured area is divided into zones, often shaped as circles or as buffer zones from the coastline. Insured zones can be ‘nested’ with multiple inner and outer zones.
- Validating Wind Speeds:
- As the cyclone moves, its maximum wind speed is measured at different track points within the predefined zones, as reported by meteorological agencies such as the Fiji Meteorological Service’s “RSMC Nadi-Tropical Cyclone Centre” or the Joint Typhoon Warning Center (JTWC). The wind speed is linearly interpolated between reported data points.
- Stronger wind speeds in inner zones are considered more damaging than the same wind speeds in outer zones and therefore trigger larger payouts.
- Applying the Parametric Loss Matrix:
- A table (called a Parametric Loss Matrix) determines the payout based on wind speed and zone. The wind speed is mapped to the tropical cyclone category from the Australian Bureau of Meteorology (BOM) scale. A payout is provided if the windspeed category within the insured zone exceeds a pre-agreed threshold. For example in the figure below:
- A Category 3 storm (i.e. 10-minute sustained winds between 118 and 159 km/h) in the “Inner 2” zone would trigger a 10% payout.
- A Category 5 storm (i.e. 10-minute sustained winds greater than 200 km/h) in the “Inner 1” zone would trigger a 100% payout.
- A table (called a Parametric Loss Matrix) determines the payout based on wind speed and zone. The wind speed is mapped to the tropical cyclone category from the Australian Bureau of Meteorology (BOM) scale. A payout is provided if the windspeed category within the insured zone exceeds a pre-agreed threshold. For example in the figure below:
- Determining the Event Payout: The highest payout percentage across all zones affected by the cyclone is used to determine the payout.
Figure 1: Tropical Cyclone Winston 2016 (blue track segments) entering the buffer zones (red, green and blue shapes) used for the SiS trigger. The maximum BOM category within the shape determines the payout amount, which in this case would trigger a payout of 100% payout of the policy limit since the Inner 1 shape includes a reported maximum wind speed corresponding to a Category 5 event on the BOM scale.
The CIPI is based on the number of people impacted by a tropical cyclone and also takes into account the greater cost of reaching impacted populations located in remote areas. The CIPI trigger is designed to capture events which may not necessarily result in a payout under the SiS trigger, as it considers all tropical cyclones in the vicinity of the insured area, not just higher category cyclones which enter the coverage zones of the SiS trigger.
- Population Disaggregation: The official population count from 2020, as developed by the New-Zealand research institute GNS Science for PCRIC using the most recent Pacific Islands censuses, is disaggregated into population points using the 30m population density map data of Meta for Good and road data from the Humanitarian OpenStreetMap Team (HOT). These population points are then aggregated into unique and distinct coverage zones. The population in remote coverage zones is increased, giving remote communities more weight to acknowledge that providing disaster relief to remote regions comes with higher logistics costs, hence to trigger larger payouts under the policy. This adjustment results in the “Remoteness Adjusted Population” for each country.
- Modelling of Cyclone Winds: Reask’s Metryc solution calculates the wind speeds of a cyclone at each specific population point, with the wind speed per coverage zone assigned as the maximum windspeed at any population point within that coverage zone.
- Measuring Impact: To allow for the fact that higher wind speeds results in greater impact on the ground, the population in each coverage zone impacted by cyclone winds is multiplied by an ‘impact factor’. Impact factors are mapped to the categories from The Bureau of Meteorology (BoM), ranging from 0.5 (for category 1 winds) to 5.5 (for category 5 winds).
- Aggregating Results: The total impacted population across all affected coverage zones is adjusted with the factors described above and summed to calculate the CIPI value for the event.
- Triggering a Payout: The CIPI value is compared to a set of pre-determined thresholds, with higher CIPI values leading to larger payouts.
This method ensures the policy reflects both the storm’s intensity and the number of people affected, including extra consideration for the increased cost associated with responding to impacted populations in remote areas.
Figure 2: CIPI Payout Calculation: Population data (1) and the Metryc wind field (2) are overlaid with the coverage zones (blue borders) (3) to identify impacted regions. The population affected in each coverage zone is calculated (4) by adjusting for remoteness and multiplying by the Impact Factor, which corresponds to the Bureau of Meteorology (BOM) category of the highest wind speed within the zone. This results in the Cyclone Impacted Population Index (5), reflecting the overall impact across the country. The Tropical Cyclone Parametric Loss Index then determines the percentage of the policy limit to be paid (6), leading to the final payout (7). Winston 2016 would have triggered a 100% payout as the TC Parametric Loss Index would have crossed 2.5M.
Additional information
Metryc: Real-Time and Historical Wind Fields
Metryc developed by Reask is an innovative solution designed to meet the demands of the parametric insurance industry and disaster response teams. By calculating wind intensity metrics with accuracy at any location, Metryc bridges observational gaps and enables rapid payouts in the wake of tropical cyclone events.
Key Features
- Hazard Intensity-Based Triggers
Metryc is specifically designed for parametric insurance, using high-resolution wind intensity data to determine payout triggers. This approach minimizes basis risk – the mismatch between actual and insured losses – ensuring equitable outcomes. - Comprehensive Global Coverage
Historically, deploying parametric solutions in many cyclone-prone areas has been challenging due to limited observational data. Metryc solves this problem by combining predictive models with publicly reported tropical cyclone data to provide accurate assessments anywhere in the world. - Accounting for Uncertainty
To account for uncertainties in storm behaviour and environmental conditions, Metryc generates 100 ensemble simulations for each event. This probabilistic approach models variations in wind field shape, terrain effects, and maximum wind speeds. The final output is a footprint map that represents the most likely wind intensities at every 1-km grid cell as presented in Figure 3.
Applications
- Parametric Insurance: Enables insurers to settle payouts quickly and fairly using predefined triggers based on wind intensity thresholds.
- Historical Risk Analysis: Supports the structuring of insurance products with data-driven insights from historical cyclone events.
- Disaster Preparedness: Equips emergency response teams with detailed hazard maps, improving resource allocation and evacuation planning immediately after an event.
Figure 3: Metryc footprint of tropical cyclone Winston hitting Fiji in 2016. This footprint is created with input data from the NADI reported track parameters using 10-minute sustained wind speeds in km/h.
DeepCyc: Probabilistic Hazard Modelling
DeepCyc developed by Reask complements Metryc by providing a probabilistic view of tropical cyclone risk. This product is particularly valuable for understanding how cyclone hazard evolves under past, current and future climate scenarios, aiding strategic planning and resilience-building efforts.
Key Features
- Physics based Probabilistic Modelling
At the core of DeepCyc is a machine learning based model that understands which climate parameters drive tropical cyclone activity. This approach ensures that DeepCyc captures the full range of possible cyclone behaviours, including extreme and rare events. - Global and Terrain-Corrected Coverage
DeepCyc incorporates local terrain features, such as topography and surface roughness, to provide precise risk assessments at a 1-km resolution. These corrections are essential for regions with complex landscapes, where wind patterns can deviate significantly from standard models. - Climate Scenario Modelling
DeepCyc provides the flexibility to analyze different scenarios, including ENSO[1] phases, as well as future climate scenarios such as +2°C global warming[2] or an emission scenario such as SSP5-8.5[3]. This makes DeepCyc an essential tool for effective climate adaptation planning.
Applications
- Pricing and Structuring: DeepCyc enables robust pricing and structuring that is consistent with Metryc, as both use the same wind field methodology.
- Climate Resilience Planning: Assists governments and organizations in preparing for future cyclone risk by offering insights into evolving hazards.
[1] https://www.climate.gov/news-features/blogs/enso/enso-and-climate-change-what-does-new-ipcc-report-say
[2] https://www.ipcc.ch/sr15/chapter/spm/
[3] https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_SPM.pdf
