Bio

By Karen Richardson

Karen Richardson is a 20 year veteran of Esri. She is a senior writer covering the use of GIS technology and data in the commercial industry and mapping and charting space.

A string of natural catastrophes over the past 18 months has cost the insurance industry billions of dollars worldwide and has forced insurers and reinsurers to reevaluate their risk management and claims response processes. Willis Re, headquartered in London, England, is a broker for reinsurance products, helping its clients – insurers – manage their risks. Willis Re also acts as an intermediary between insurers and other reinsurance companies.

Following the magnitude 9.0 earthquake that rocked Japan on March 11, 2011, and set off a tsunami, Willis Re quickly went to work gathering critical information for their insurance clients. Using a system called eNCOMPASS Online based on Esri ArcGIS software, Willis Re was able to assess areas of tsunami inundation and, using their understanding of their insurance clients’ portfolios of exposure, to estimate potential insured losses. Willis Re’s insurance clients in Japan and other territories across the globe are now giving direct access to this system to interact with their own data themselves.

Preparing for the Worst

Willis Re’s core focus is to provide insurance companies with a superior understanding of the risks they face, helping them manage extremes. When the organization anticipated a more active tropical storm season during 2011, they built eNCOMPASS Online within eight weeks to estimate the potential impact of large tropical storms on insured portfolios on behalf of their clients. Both client users and Willis Re analysts can view and discuss a portfolio by logging on to eNCOMPASS Online from anywhere in the world. The spatial analysis component of the system allows them to identify exposures or individual policies that are physically located in the projected path of an active storm and to determine the associated projected wind speed, storm surge or similar information.

“Willis Re and client users are able to see events unfold at the same time, which encourages collaboration,” said Nigel Davis, Executive Director of Product Development, Willis Global Analytics. “Discussions can take place around the levels of insured value identified within a tsunami-affected area, or can identify insurance policies that are about to be affected by a hurricane.”

Once the affected exposures are identified, all the descriptive information associated with those exposures can be exported for further analysis as required outside of the system. For example, an insurer could apply some loss calculations, plan emergency operations or even contact policy holders ensuring that response and customer service are accurate and timely.

Willis Re consumes or develops live feeds for real-time or near real-time information on events. For example, a process to download United States Geological Survey (USGS) data was developed by Willis Re to display recent significant earthquake activity across the globe. The platform does not contain just active hazards, but also historical events and risk models and indices. In particular, it is becoming a window for research outputs of the Willis Research Network – the world’s largest partnership between academia and the insurance industry, in which Willis Re has teamed up more than 50 leading universities and institutions worldwide who are involved in atmospheric science and climate statistics, geography, hydrology and seismology. Through this unique collaboration, insurance client users are increasingly going to be able to interact with leading research in conjunction with their portfolios.

Willis Re relies heavily on good quality information about risk exposures and the ability to communicate the results of its analysis in an efficient, comprehensive way. Clients can then make better decisions under the elements of uncertainty that are associated with this risk. “eNCOMPASS Online as a delivery platform gives us and our clients the means to visually understand exposures and the benefits of better information,” said Enrique Venegas, a leading Latin American broker with Willis Re. “For us to be creative and effective in this way allows us to deliver and convey the critical outcomes of our analyses.”

The Data Feeds Needed

When the earthquake and subsequent tsunami hit Japan, Willis Re staff derived a bespoke (or custom made) tsunami zone using GIS geoprocessing, a digital elevation model (DEM) and ground observations in order to create an estimated representation of the affected areas. The DEM was derived from Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensors. ASTER provides remotely sensed terrain data that is easy to access, a reasonable resolution (30m) and a wide coverage. Elevation and slope of the land were derived to analyze where inundation from water would take place.

The earthquake shakemaps from USGS on top of a world topographic map from Esri extended the data available after the event. “Using eNCOMPASS Online, not only are we now able to offer event response capabilities to our Caribbean and Asian clients that we weren’t able to before; we can also provide these services in record time,” said Davis.

Finding the best representation of an insurer’s exposure can often be difficult. In many cases, the broker receives exposure data that is reported by administrative boundaries rather than receiving accurate street address latitudes and longitudes for each insured location at the street address level. This problem can be the product of incomplete data capture, poor addressing systems in a particular country, or perhaps a high investment cost for geocoding to high resolution. Willis Re has developed disaggregation techniques to produce an artificial but more accurate representation of exposure than simply working with data geo-referenced by postcode or municipality centroids. This system aims to lessen spatial uncertainty and inaccuracies associated with using centroids of large areas.

While the cost and effort of collecting and managing higher resolution exposure may sometimes seem unattractive, the value can quickly be realized in terms of impact on CAT model losses or changing damage estimations after an event. This is particularly true when dealing with hazards with very localized effects.

After the various catastrophes this year, Willis Re observed an increase in the appetite to collect better address data to use in tools such as eNCOMPASS Online to facilitate decision making. The value of higher resolution postcodes and street addresses has been recognized and perhaps better quality data will be passed up to the broker. “Unfortunately we have seen that a disaster tends to emphasize the importance of having better data,” said Davis.

Crowdsourcing: Remote Sensing Versus Near Sensing

Another aspect that is assisting in more accurate data is a phenomenon called volunteered GIS information (VGI). “Suddenly, we have all of these people on the ground who are connected by virtue of their personal devices – laptops, smart phones, digital assistants – you name it,” said Davis. “Instead of remotely sensed data, we are getting ‘near sensed’ data – from people who are using social media to report on events as they happen via text messages or photos they take on the ground.”

If these on-the-scene reports contain a location – and many do – this information can be harnessed. Location-aware social networks have a huge potential for enabling people in a community to help themselves during a crisis. Spatial analysis makes crowdsourced data actionable. While one Flickr, Twitter or Facebook post may not be critical, if there are many posts, a heat map from data collected on the ground pertaining to property damage, hazards, evacuations, power outages, and help and services can be collected from the best source of data – those affected.

Location-aware social networks have a huge potential for enabling people in a community to help themselves during a crisis. Spatial analysis makes crowdsourced data actionable.

Visualizing this data as hot spots or trends could give an idea of the density of reports coming from a single area. This might assist in allocating resources to those who need the help most or aid the validation of whether incidents reported are corroborated by others, adding increased clarity to a situation.

More than just dots on maps, hot spots organize lots of data and provide a better understanding of the data quickly. Reports can also be filtered by date to see daily trends. Providing this spatial information in a spatial context can connect individuals and optimize the use of trained resources. These areas are being explored further as Willis develops its understanding and use of social media within an insurance context.

“GIS can help brokers find all the Flickr posts within 100 meters of an insured property, for example, to get real understanding of damage and claims from those actually in the area,” said Davis.

Continuous Innovation

The Internet is a natural platform for geospatial analysis. Many participants can move from data sharing to creating shared services easily in this environment. The technology makes it possible: collaborative computing, service integration, mash-ups, user-contributed content and distributed data management are the many ways that access has been opened to many new users and applications.

The reinsurance broker continues to push the envelope by offering personalization of information for clients, such as loading only certain risks based upon a client’s zone, and developing interfaces to parts of the application through mobile devices such as iPHONE and iPAD. “Willis Re is one of the first to understand and implement a solution that integrates all levels and supports open access, collaboration and transparency,” said Mark McCoy, insurance industry manager at Esri. “They are able to do this because they have access to authoritative data, and can create high quality maps that support visualization and spatial analysis and models through a rich application.”