Underwriting and remote sensing: Satellites to selfiesBy Ron Richardson and Steph Bednar | February 1, 2017
Insurers need the most accurate and reliable risk information available to underwrite a policy. An emerging source of this data is remote sensing technology, which is the use of sensors and cameras on space-borne, airborne, and ground platforms. The images help insurers understand risk factors of properties and surrounding areas, weather effects, and many other details.
Most important, the data and analytics provide location-based, property-specific ground truth information. Remote sensing technology helps insurers capture property characteristics, detect property changes, assess property liability risks, estimate damage, and much more.
'True diagnostic approach'
A key value of the technology is that it requires little human assistance to capture data from the image. The data can feed into information dashboards and help create analytics that insurers can use in a variety of ways—a true diagnostic approach to underwriting.
The main sources for high-resolution images are satellites, airplanes, drones, and mobile devices. Each has distinct advantages and limitations, and, working together, they give insurers the most accurate and reliable ground truth data possible.
For the purpose of this article, when we talk about resolution in remote sensing, ground sample distance (GSD) in a digital photo of the ground from air or space is the distance between pixel centers measured on the ground. For example, in an image with a 3" GSD, adjacent pixels image locations are 3 inches apart on the ground.
Satellite imagery offers a unique high-level view of a property. Image resolution is lower than aerial but can reveal risk details and key data points about properties to underwriters, and updates can be delivered with high frequency.
Current satellites can map Earth's entire surface up to 12 times per year. Such frequency allows analysts to confirm information gathered from older, higher-resolution images and observe changes to properties over time. Insurers can see how weather, property damage, exterior renovations, or property additions (a pool or new structure, for example) have changed risk factors. Satellites can cover broad geographical areas but aren't as efficient for detailed, close-up property analysis due to their lower-resolution images.
Satellite imagery resolution continues to improve. In the 1980s, satellite image resolution exceeded 30 feet. Today, commercially available satellite images are as granular as 1-foot resolution, detecting a wide variety of features such as central air-conditioning units, roof cover, and roof shape.
Micro satellites, some as small as a tissue box, are another advancement. Less expensive and easily deployed, they can capture a single point many times a day. Users can plan the target and frequency, allowing for greater and more specific analysis in tracking wildfires, floods, severe weather, and other events. Image resolution for micro satellites typically exceeds 6 feet.
Imagery from low-flying, airplane-mounted cameras and sensors can provide much higher resolution than a satellite (less than 2-inch resolution), letting insurers glean much more information about a property.
The aerial platform allows for the capture of oblique images because planes can take pictures at 45-degree angles in addition to top-down, or orthogonal, images. Obliques are vital for measuring roof slope, elevation heights, and window and door sizes and for providing information on extra surfaces, new structures, and other changes. Obliques create true 3D property models.
Resolution of aerial imagery is constantly improving and becoming more varied. Five years ago, the technology of aerial cameras was only able to provide 12 to 16 megapixels. Today, with advancements and improvements in technology and dynamic range, aerial cameras provide up to 80 megapixels, and 160 to 280 megapixel images are coming.
In addition to cameras, the aerial platform features sensors that perform a variety of tasks. Using laser technology, sensors can measure distances between objects, such as properties or surface features for terrain mapping. Multispectral and hyperspectral sensors can capture light frequencies the human eye can't see that can be used for material detection. Infrared and ultraviolet sensors use colors to identify living and nonliving matter.
The data and images feed directly into our databases and online systems to help create baseline 3D property models or a profile of a property.
Drone use has exploded in the last few years, and the commercial market in the United States doubles every year. More companies are learning to deploy drones to great advantage, and that includes insurers. Drones flown closely to a property can capture high-resolution images from multiple angles and target specific features.
Drones can be relatively inexpensive and equipped with light or heavier cameras depending on the type of drone, giving them even more versatility. The types of drones are diverse-rotors, winged, fixed-blades, flying copters, gas-powered, battery-powered, and more. Some units are small enough to fit in your hand; others need a truck to carry them. Drones continue to evolve and become more specialized. Data and imagery from drones can also be used to produce 3D property models.
Drones can take thousands of pictures of all types of terrain to determine anomalies. Soon, data and images captured from drones and combined with estimating software may be able to help detect hail and wind damage for a claim, assisting insurers in writing structure damage estimates.
The second-most popular use of a cell phone, after searching the Internet, is to take selfies. Insurers can take advantage of that proclivity because technologically advanced smartphone cameras can take efficient, inexpensive, and comprehensive ground-level images. Mobile devices are just another place to put a sensor. Most people have them anyway, so there's no additional cost for the technology.
Cameras on today's mobile devices can capture high-resolution images and data as an additional method of providing property details. A person can take pictures all around a property for a 360-degree view. Combined with satellite, aerial, and drone images, you can create an extremely accurate profile of the property.
Using state-of-the-art computer vision techniques, it's possible to pinpoint angles and surface details down to a highly granular level. Tracking functions can reveal the type of sensor used, where the picture was taken, and the direction the photo was taken. And you can link to metadata from a compass, GPS, or other application.
Use case benefits
Each remote sensing technology has its advantages for insurers. Satellite images can provide current information and measure changes to a property with high frequency. The better resolution and oblique images from aerial imagery establish important baseline data, reduce reliance on third-party input, and reveal the true risks of construction factors. Drone imagery can be used to survey specific properties for underwriting and claims. Mobile devices offer immediacy and detail, and images can be taken from any angle, making it easy to see changes from prior inspections.
To get the most benefit out of remote sensing technology, however, insurers can use these imagery-based solutions together across their entire organization, from point of sale to claims. Combining imagery from all platforms provides a virtual inspection that's reliable and as accurate as possible. The information is powerful throughout the policy's life cycle, and insurers can use it proactively to analyze portfolios, research potential customers, and prepare structural damage repair estimates. And you can accomplish all this without additional on-site inspections or potentially intrusive customer touch points, further reducing costs.
Enhancing ground truth data
Remote sensing and imagery-based data allow insurers to get to the real ground truth of a property, especially when all the different sources of images work in conjunction to provide a detailed, highly accurate risk profile. Ground truth data reveals the actual, current property condition and indicates changes from previous inspections. It improves upon the information derived from other data sources, such as public records, tax assessor data, on-site inspections, and policyholder input. Capitalizing on ground truth data makes underwriting and rating a property or settling a claim as accurate as possible, inevitably helping to lower costs, protect policyholders better, and increase profitability.
Ron Richardson is assistant vice president at Geomni, a Verisk business specializing in technologies related to image capture and analysis. You can contact Ron at email@example.com. Steph Bednar is a product manager with Verisk – insurance solutions. She can be contacted at firstname.lastname@example.org.
Tagged with: ,