J. Arn Womble

Automated Building Damage Assessment Using Remote- Sensing Imagery

The automated comparison of before-and-after remote-sensing imagery provides an effective means for rapid and widespread assessment of windstorm damage to individual buildings. The development of automated damage-assessment algorithms involves the classification of building damage signatures from a remote-sensing perspective, the identification of corresponding temporal change metrics, and the correlation of remote-sensing change signatures with actual field-based damage observations. Hurricanes Charley (August 2004) and Ivan (September 2004) marked the first major hurricanes for which high-resolution satellite images were available. These storms provided an exceptional set of before-and- after images. Investigators from Texas Tech University and ImageCat, Inc. obtained temporal satellite image sequences and performed associated ground-truthing damage surveys for these major hurricanes. This paper chronicles the use of the before-and-after hurricane imagery to develop remote-sensing-based damage scales for various building inventories; the correlation of remote-sensing damage metrics with field-based damage investigations; and the progress in automated damage assessment using temporal image sequences.

The Enhanced Fujita Scale: For Use Beyond Tornadoes?

The Enhanced Fujita (EF) Scale was developed specifically for the estimation of wind speeds based on damage caused by tornado winds. Recently, the question has arisen as to whether or not the EF Scale can be reliably used in reverse to predict the amount of wind damage based on measured or estimated hurricane wind speeds, particularly when wind action is co-mingled with severe storm-surge action and evidence of the exact level of wind damage is obliterated. In considering such use of the EF Scale, we examine separately its two major components: (1) Degrees of Damage and (2) ranges of tornado wind speeds associated with these damage levels. Our experience suggests that, in general, the EF Scale properly describes the overall progression of damage with increasing wind speeds in both tornadoes and hurricanes; however, the wind speeds associated with various degrees of damage are generally expected to be different for structures exposed to hurricanes and tornadoes. Use of the EF Scale to predict hurricane damage based on peak wind speeds is therefore discouraged. DEVELOPMENT AND USE OF THE EF SCALE The original Fujita Scale (Fujita, 1971) was designed as a wind speed scale to provide relative ratings of tornado intensity (McDonald et al., 2009). The scale divided intensity into multiple wind speed ranges and attempted to describe the type of damage one might expect for each wind speed range. This original Fujita Scale did not account for differences in the resistance of various structures. Prior to the 1970s, there was little understanding of the true magnitudes of maximum wind speeds in tornadoes due to a scarcity of tornado wind speed measurements. Engineeringoriented investigations of tornado damage subsequently highlighted the importance of

Common Misconceptions in Determining Wind/Water Damage Causation

Recent severe tropical storms such as Hurricanes Katrina (2005) and Ike (2008) have resulted in questions of whether wind or water (storm surge) is responsible for severely damaging or destroying structures along the coast. Because of a separation of insurance for wind and storm-surge (flood) damages in the U.S., the determination of whether hurricane damages are the result of wind or of storm surge becomes a crucial issue in the settlement of insurance claims. Disagreements as to the cause of such damages have led to an unprecedented series of legal challenges between homeowners and insurers. Engineering and meteorological reports directed at the performance of residential buildings have been prepared in support of both sides of this issue. Many such reports employ failure theories that are implausible and that can be potentially misleading. The failure theories in question appear to stem primarily from a misunderstanding of wind and storm-surge loading of structures, from a misunderstanding of the behavior of structures under extreme loads, from selective or incomplete use of physical principles, from the use of incorrect or non-standardized wind and surge data, and/or from incomplete analysis of clues provided by the damaged structure and surroundings. Physical principles that are mistakenly or inaccurately applied include fluid-structure interaction, buoyancy, and transport of structures and debris. Among the data improperly employed in recent analyses are unverified and non-standardized wind speeds and data from preliminary (non-calibrated) numerical models. Improper application of the EF-Scale relating tornado damage to wind speed ranges to hurricane damage also has been frequently observed. The supposition that tornado damage is responsible for the selective destruction of structures has also been mistakenly used in place of considering variations in structural resistance. Our experience has prompted the following recounting of physical principles, anecdotes, observations, and recommendations. This paper examines common misconceptions and provides guidance for application of the appropriate physical principles.

The Enhanced Fujita Scale: Development and Implementation

The original Fujita Scale (F Scale) was introduced as a means of estimating the intensity of tornadoes and hurricanes by relating appearance of damage to wind speeds. The relationships were based on the extensive experience of Dr. Fujita but were never independently verified. Furthermore, the importance of construction quality was not recognized, and the number of damage indicators was limited. The Wind Science and Engineering group at Texas Tech University agreed to lead an effort to improve the original scale. The improved scale, known as the Enhanced Fujita Scale (EF Scale), defines 28 damage indicators and a number of degrees of damage for each one. A process of expert elicitation was used to relate estimates of expected, upper and lower bound wind speeds for each degree of damage. The Enhanced Fujita Scale was then correlated with the Fujita Scale by means of a regression equation. The National Weather Service implemented the Enhanced scale in February 2007. In its present form, the EF Scale is applicable only to tornado intensity.

Constructing Probable Wind and Water Damage Sequences from Timelines — The Technical Perspective

Hurricane Katrina (2005) brought unrivaled destruction to buildings along the Mississippi Coast, generating thousands of property insurance claims. Hurricane Katrina had been a Saffir-Simpson Category 5 storm only 18 hours prior to making landfall at the Category 3 level along the Louisiana/Mississippi state line. At landfall, it still contained enormous potential for storm surge — far greater than the Category 3 landfall designation (based only on sustained windspeed) might indicate. As a result, damage from storm surge tended to be far more severe than damage from wind action, though many buildings sustained some levels of damage from both storm surge and wind. Exclusion of flood damage from standard property insurance policies has resulted in partial or full denial of many Hurricane Katrina-related damage claims, as District Court rulings in Mississippi have considered flood damage to include damage from storm surge. An unparalleled series of legal challenges has followed — pitting homeowners against insurers and requiring resolution of such questions as: 1 Was the residence destroyed by wind or water? 2 How much wind damage, if any, was sustained prior to the eventual destruction of the residence by storm surge? Settlement of these legal challenges has necessitated the separation of probable wind damage and storm-surge damage. The separation of damages is particularly complicated when structures have been completely destroyed or when engineers have been asked to determine wind and water damages long after critical evidence has been removed from the site. However, additional significant data concerning region-wide wind-speed and storm-surge levels (and the timing thereof) may not be available until several months after the hurricane, once detailed analysis of available region-wide observations has concluded. These regional timeline data provide a valuable resource for engineers in the reconstruction of probable wind/surge damage sequences some time after the hurricane damage has been removed.

Imagery-Based Wind Damage Functions

Prediction or estimation of wind damage to individual structures using analytical techniques is time-consuming, costly, and often limited by unknown loading factors and structural resistances. Detailed analysis of all individual structures affected by hurricanes or large tornadoes is typically not feasible. Alternatively, wind damage functions (WDFs) for structural components and/or overall systems based on the observation of damages distributed spatially across a known wind field can serve as a significant tool for the estimation or prediction of wind-induced damages. The intensity and spatial distribution of maximum wind speeds in severe wind storms have historically been unknown, and the collection of area-wide and comprehensive damage data via ground surveys has generally proved impossible. However, it is now possible to construct improved wind damage functions based on high-resolution post-storm imagery by correlating observed damage levels and known wind speeds established through the merger of significant new technological developments, including: portable ruggedized instruments that measure wind speeds in-situ, numerical modeling techniques that produce regional velocity fields from available multi-platform measurements, remote-sensing platforms that facilitate the rapid capture and preservation of damage data, and interpretation techniques that aid in identifying levels of wind damage depicted in these remote-sensing data.