"Although we have always thought data was important in an emergency, I now think probably it is second only to the first responders’ courage."

Information to support the response was a critical concern. For everyone, information needed to be relevant, sufficient, accurate, timely, and accessible. For first responders, it also needed to be constantly updated and usable by people in extremely hazardous conditions. None of the information resources used in the response and recovery had all of these qualities, and many compensating compromises were made to get information into the hands of those who needed it in a form that would help and not hinder their activities or safety.

"What you need to be prepared for in an emergency, what you have to have prepared, is the data elements that go into supplying field forces with the data they need to respond."

Relevant information, that is information suited to its intended use, was needed to support emergency and recovery operations of all kinds. Ideally, this information would be organized in a way that is easily accessible and readily usable by those who need it. However, many interviewees told us they lacked a useful understanding of critical information systems and data sets that would be relevant to either immediate response or long-term recovery. This understanding was developed in a limited way on the spot and the result was often frustrating and fraught with gaps and inadequacies. Available information was often out of date, at the wrong level of detail, or incomplete. Analysts and managers spent a great deal of time and energy trying to evaluate and improve the relevance of these data resources but, as one interviewee observed, "a crisis is not a time to be making decisions about data processes."

Relevant information about some topics could not have been acquired in advance because the information pertained to the event itself. For example, the need for aerial imagery to assess the extent of damage and understand how to approach the rescue effort and fires on the debris pile at Ground Zero demanded that new data be captured. This was accomplished by using fly overs of the site with both standard and advanced forms of photography and remote sensing. In this case, the data could not have been gathered in advance. However, interviewees noted that knowing in advance how to acquire and distribute such information would have been of great value.

In addition, relevance was an ever-evolving concern because the nature and priority of information needs changed over the course of the response and recovery. While immediate emergency response required information to support rescue activities and deal with imminent danger, later work depended on detailed information about equipment, organizations and individuals on the site, victim identification, service offerings and locations, buildings in the surrounding area, public health threats, and other topics.

Interviewees offered several strong recommendations for increasing the amount and usability of relevant information. They emphasized the need for all organizations to create and maintain complete inventories of data resources and information systems. Had these existed, emergency managers and supporting organizations would have been able to direct their attention immediately to data analysis, rather than to data search and evaluation. This was not at all the case in the WTC response. Even routine hazard information, such as chemical fact sheets, was unavailable and this lack of information contributed unnecessary uncertainties for the response and recovery workers about the conditions of the site and surrounding area.

Others described how prior agreements about predefined initial information products, including standard reports and maps, would have aided both decision makers and the public. Lacking these, multiple versions of reports and analysis were invented as the event unfolded. According to one respondent, "There [were] problems ... collecting . . . data on the site because none of the templates and the formats had been prepared ahead of time, none of the coordination between the gathering bodies had been worked out ... It all had to be done on the fly." Respondents said that basic decisions about report topics, frequency, formatting, size, color-coding and so on could all be made in advance. While these pre-defined information products would surely need to be amended to fit a particular event, they would form a sound foundation of well-understood relevant information for a variety of users and uses.

"[At] different points there was certain utility data that we needed to get, data that was considered reasonably sensitive by the utilities... We did get it but it’s scary to find out that the data’s just in paper form and we ended up having to digitize it ourselves. . ."

In some cases, needed information existed but was not accessible or in usable form. In others, the amount of information was simply inadequate. As a result, emergency crews had almost no usable data during the first 24 hours after the event. For example, New York City officials estimated that only 25 percent of the information that emergency crews needed existed before the event, and most of that was lost, at least temporarily, as a result of the attack. Most of the first week was spent rebuilding pre-existing data capacity. Interviewees recommended that response planning include special attention to building and testing various event scenarios specifically to determine what data would be sufficient for a first response and how it could be organized and delivered in a timely manner.

At the state level, interviewees believed most of the data they needed already existed, but it was not always readily available or usable. They sometimes lacked written procedures for key activities associated with data selection and management, and some of the data they had lacked the detail necessary to make it immediately useful. For example, contact lists of technology suppliers were used extensively, but often contained little more than company name, address, and general categories of products. Today, those same directories are being revised to include the identity of specific individuals to be contacted in emergencies and details about products and categories of expertise that each business might bring to a new crisis.

Other essential information existed but could not be readily accessed or used. For example, the WTC site was the location of a number of hazardous materials. These had been cataloged years before, but this information existed only on paper and was located in Albany. It took days to find and make use of the information. Once it was located, the information turned out to be too general to be very helpful. Sometimes valuable information necessary for re-establishing normal operations in non-emergency organizations had been kept only on paper, such as legal files for cases in the process of litigation. This information was either destroyed or made inaccessible due to the closure of buildings that needed thorough inspection or repair before they could be re-occupied.

Nevertheless, critical information replicated in different locations did allow for quick recovery of most City services and large businesses. Thanks to planning for the Year 2000 (Y2K) date change, and to standard practices for disaster recovery, most organizations had back-up files of key data. However, back- ups were not always kept in remote locations and sometimes had to be retrieved from quarantined buildings. Sometimes duplicate data sets were not the result of planned back-up procedures or business continuity plans. Rather, data were sometimes replicated just for ease of use, to alleviate bandwidth problems, or for other business reasons. Perhaps the best known instance of this kind of duplicate data was the City's GIS base map, NYCMap, which was replicated at Hunter College under a contract for quality control work. When the EOC collapsed and the City's IT office in lower Manhattan was closed for safety reasons, the files and associated hardware and software at Hunter College allowed for the quick and effective re-establishment of GIS data services.

"We got data from the state about underground fuel storage and freon and it turned out that this was very critical data, but the way it was captured was very vague as to exactly where it was . . . We spent days and days trying to figure out the exact location."

Accuracy is a combination of both factual correctness and appropriate level of detail. A NYC official who was involved in preparing data about structural features, such as shafts and fuel storage facilities under the Trade Center, indicated that the existing data was vague and ambiguous, thereby rendering it mostly unusable by first responders. This kind of experience highlighted for many the need to develop and consistently employ data standards to guide routine collection, storage, updating, and delivery of building data that is accurate at various levels of detail from building footprints to wiring diagrams.

One NYC official who worked at Pier 92 told us about the earliest efforts to prepare data for responders. ". . . first of all in terms of accuracy, nobody would stand behind any of it. And that was all very clear . . . all that was being mapped to just try to get some picture of what was there and to give the first responders a leg up in terms of possible voids, maybe where there were shafts and that kind of thing. It was mostly for them to use on the site to prevent their own accident. You would not want to do that on the fly again; that's why I'm probably going to dedicate the rest of my career to getting this data now and having it, just in case."

Moreover, several respondents discussed the importance of documenting data characteristics through meta data that states ownership, provenance, definitions, limitations, data collection method, and timing, and so on. Very little meta data was available to help analysts and users understand the appropriateness and reliability of data for various uses. In addition, little attention could be paid to records management or archiving because no protocols had been set up in advance. These were low-priority activities in the first few days, although they received increasing attention as data analysis activities expanded.

"You may normally want to get 95% or 90% complete data, but it may be better to get 75% in some cases but get it out so the people can use it."

Relevant and accurate data that is not available when needed fails to serve much purpose. As the response activities got underway a variety of new data sets were contributed by or collected from both public and private organizations. Processing and delivering this information to first responders and rescue planners in a timely manner took on extreme importance. Although everyone was concerned that incomplete information could be harmful in some way, the pressure to get information into the field was intense. One respondent commented that it is sometimes better to collect less data if it would be more usable and able to be communicated to others faster. "You don’t want to give anybody bad data or bad analysis or anything like that but if the color’s not quite right or the legend isn’t right or whatever, that is secondary to getting the information out."

One interviewee recalled a debriefing meeting in November 2001 in which the fly overs were discussed: ". . . this digital ortho imagery, it was pretty nice and . . . you got it down to six or eight hours and it was driven down to the city . . . [But] it wasn't good enough . . . it was good for what we needed for other things but it wasn't good enough [for those first days]. I need something in two to three hours. And in fact, we took somebody with a digital camera and put them in a police helicopter to film the fire every day 'cause we just needed to see a picture of where it was. We used the other stuff . . . but we need something else; we need something much quicker."

"The spatial element enables data that you have to be brought to that higher level [of usefulness] by being able to integrate it, analyze it, and present it in ways that you just can’t do conventionally."

Interviewees pointed out that an emergency is almost always a spatial event. Consequently, mapping and geographic data analysis were crucial to response and recovery efforts, and to providing public information. The visual aspects of spatial data make it remarkably versatile and suitable for a wide variety of audiences, including expert analysts, emergency response teams, policy makers, and citizens. The use of maps to convey status and safety information is one of the unquestionable success stories of the response.

Spatial data provides a comprehensive view of many different attributes of the geographical region impacted by the emergency such as physical geography, critical infrastructure, building footprints, transportation routes, and demographic characteristics. These types of information can be geo-coded (associated with an exact place) which then allows them to be combined, compared, correlated, or integrated to produce new information. According to one participant, the experience of using spatial data "led us to profoundly understand the importance of place and location and organizing data according to spatial attributes because then the data makes more sense from it being combined."

Geographic information systems (GIS) and location-based information services on the Web emerged as the most versatile analytical tool associated with the response. Fortunately, NYC had been developing its GIS base map, NYCMap, since 1998. It was created by combining aerial photos with limited information about the City’s physical geography and built environment at both the surface and subsurface levels, all accurate to within 18 inches. Within a day of the event, an Emergency Mapping and Data Center was created within the EOC, drawing data from many different sources, and delivering more than 7000 maps to first responders, public safety authorities, utilities, the media, and others. These maps showed critical and ever-changing information in a readily usable way. They depicted a wide range of information from thermal photography to pinpoint underground fires for emergency workers to subway maps showing the public which lines had been restored to service.

These GIS applications made use of the City's existing investment in digital mapping and data sets, integrated with other GIS data sources pulled from many other organizations. This experience highlighted the importance of having readily available digital maps and layouts of facilities and buildings as a part of the emergency recovery templates. When they can be combined with dynamic event data these maps can become powerful tools in the hands of first responders.

However, as well as GIS performed in the WTC disaster, the spatial analysis team was far from satisfied with what they were able to do. There was no single, authoritative repository or directory of relevant and reliable geographic data that could be immediately deployed. Complete information about utilities and other critical infrastructure was not part of the existing GIS system and was not fully integrated when acquired. In addition, interviewees who conducted spatial analysis said the lack of meta data about the definitions, sources, and usability of the data forced unacceptable delays while they tried to integrate different data sources into something that was usable in the field. Finally, detail-level data for most buildings that described such characteristics as construction materials and uses simply did not exist.

"Public safety professionals have years of field experience but don’t have much experience manipulating data. We need to figure out ways to educate them and to make sure that they understand the technology so they find it usable."

A variety of new technologies were made available as part of the response. However, a number of interviewees discussed the difficulties of introducing new information management and analysis tools in the midst of the crisis to first responders who are much more attuned to direct observation and action. During a crisis such as the WTC, first responders are required to make decisions in an environment characterized by extreme stress, danger, confusion, and uncertainty. To be effective in such a state, data for decision making should be presented in a familiar form that does not engender any additional cognitive burden on first responders.

Consequently, we heard strong recommendations for ongoing orientation, training, and practice using new technologies. This will entail finding ways to help first responders, decision makers, and public affairs officers develop an appreciation for the abilities and limitations of data-driven action. Interviewees stressed a consistent list of necessary but currently underdeveloped or entirely missing information strategies. These include organizing data for easy access, using well-understood templates for presenting information to various audiences, pre-designing basic maps using common intuitive symbols, developing and practicing familiar online applications, and adopting uniform data standards. In addition, knowing in advance what kinds of data to bring to a response would go a long way to ensure the right data is used by first responders leading to better outcomes.

During the following winter, prompted by the experiences in the WTC response, an industry consortium called OpenGIS, in cooperation with a variety of government agencies, began to explore a test bed for real time collection, reporting, and integration of remote sensing data over the Internet. Called SensorWeb, it represents a first effort to explore the ways in which advanced technologies can support the information needs of first responders.

"I saw people scrambling to put an information-sharing capacity into place . . . but you can’t stop there . . . you really have to dig into the science of who’s collecting what data, where, when and why before you can assume that you can throw it all together on a screen and put it out to people."

Data coordination and integration problems quickly surfaced after the attack, persisted throughout the response, and continue into the present. The lack of data standards (and lack of policies about data standards, sharing, and coordination) took a significant toll on the response effort. Disagreements among the various levels of government about whose data was most accurate or most suitable for different applications cost precious time in the early days of the response. For example, FEMA has spatial data that covers the entire nation at a standard scale, but local spatial data, while not always standardized, is often far more detailed and suitable for supporting on-the-ground operations. One respondent described how the physical proximity of staff from various agencies on Pier 92 helped mitigate some of these problems. The agencies represented in the EOC needed maps that required information from different sources be brought together. This necessitated decisions about whose data was best for a given purpose. Because so many organizations shared work space on the Pier, they could deal with this problem face to face, "so that resulted in standing around [in a group] and getting people to make a decision as to whose data was better. That data cleaning and integration service was very informal but ended up being the basis for a lot of trust in the product."

Nevertheless, many data sharing problems remained. Environmental health data, a source of major controversy both during and after the event, is collected at every level of government, but there are no agreed upon standards or divisions of labor that make these data resources fit for immediate integration and coordinated use. According to one respondent, "Our problem was that we had everybody doing sampling; some of them were doing sampling in the same way; some of them were doing sampling in a different way; some of them were representing the data the same and some of them weren't." Instead of having data resources that complemented one another, the environmental protection agencies were faced with either competing data or gaps where no one had collected information before.

Moreover, intergovernmental and internal agency conflicts developed around the nature and form of information to be released to the public about environmental health risks. Both political and scientific concerns caused delays and resulted in little useful information being disseminated in the early days of the crisis. In this case, the City and federal environmental protection agencies were involved in monitoring and collecting two different types of local environmental data (asbestos levels and air quality levels). It was essential to integrate these two sets of data with the local GIS data in order to make dynamic zoning decisions that affected the movement of citizens and relief workers. However, this was extraordinarily difficult due to the nature and quality of the data and the lack of agreement about how to report it.

One respondent spoke at length about the importance of data standards and the difficulties associated with not having them. "What I was watching [were] folks from different agencies at all levels of government. . . having a complete inability to relate their data to one another. Neither in terms of who was collecting it and when it was collected, how it was being collected, how it was being analyzed, what it meant, definitions, terms . . . the whole thing was really quite unfortunate. . . It’s a major, major issue that affects our ability to respond in emergencies, but it also affects our ability to optimize resources and share information and make good investments."

This problem was evident in a variety of situations. For example, multiple lists of the dead and missing compiled by different organizations needed continually to be reconciled. Multiple addresses and names for buildings was another important data problem. Buildings in lower Manhattan were inspected and re-inspected by six different government agencies before they could be declared safe and reoccupied for residential or business use. The lack of a single accepted building identifier slowed and complicated the process for all concerned. One interviewee described this issue: "The primary difficulty was not knowing the total story of any one facility. So a facility that has multiple floors, data centers, different connectivity points, different vendors supplying connectivity to that building, you couldn't know at any one point in time, is this floor up? Is this floor down? Do we have a connection? Do we not? All of that trickled in as people started touring the buildings and finding out what was working and what wasn't." This experience prompted the City government to embark on a building identification program to reduce or prevent such problems in the future.