1. Introduction: The Need for Disaster Risk Management Data

Understanding Risk

Disasters reveal chains of decisions about risk (Natural Hazards, UnNatural Disasters (NHUD), 2011). When infrastructure fails under strain of an earthquake, citizens may point to the failure of the construction firm to adhere to building standards. Or to the failure of a government to set and enforce code around retrofitting a school to seismic risks. Or to the owner of a factory to have inquired into the exposure of the structure to hazards and developed a strategy to cope with these potential vulnerabilities.

In each case, critical information is missing. Information that might have driven a different choice about architectural designs, building materials, or the site for the building (siting). Information that might have driven a community to question choices. Information that might have driven a legislature to pass laws or officials to allocate staff time to enforcing them.

Credible information about risk is an essential element of Disaster Risk Management (DRM). Across the disaster risk management cycle, institutions are now engaged in a process to build this stock of information. The aim is to improve the chain of decision across entire system, from the donor who funds retrofitting of schools to the individual business person who need to mitigate potential losses from natural hazards while caring for his or her household.

---

Risk Assessment and Communication

Building risk assessments follows an internationally agreed approach that combines three elements: hazard, exposure and vulnerability.

• Hazard: The likelihood (probability/chance) and intensity of a potentially destructive natural phenomenon. For example, ground shaking in an earthquake; severe wind/storm surge in a cyclone; inundation level and velocity in a flood event.
• Exposure: The location, attributes and value of assets that are important to communities (people, property, infrastructure, agriculture/industry etc).
• Vulnerability: The likelihood that assets will be damaged or destroyed when exposed to a hazard event. For example, the relationship between the intensity of earthquake shaking correspond and the level of damage for different buildings (Figure 2).

The iterative process of understanding the threat from a given hazard unfolds through several levels of complexity. At the most basic, an analyst can model the impact of hazard estimates what might happen to people, buildings, crops etc from a single event, such as 1/100 year flood or an 8.1M earthquake. This is Impact Modeling.

Impact modeling is not the same as risk assessment. Risk is the composite of the impacts of all potential events (e.g., using looking at the impact from 10,000 cyclone events); this allows an agency to determine the annual average loss and probable maximum loss from individual or multiple hazards. Risk models can be very useful for financial planning and protection, prioritizing DRM investment within a country, and cost-benefit analysis of different risk reduction options. They are the basis for projects that build preparedness, focus risk reduction investment and action, and implement policies that slow the creation of new risks.

Managing Risks Requires Managing Risk Data

Effective disaster risk management requires a commitment to collect, curate, and analyze data over the long term. However, many governments have lacked the resources to be stewards of the data that are essential for risk assessment. Many more lack the capacity to turn the data into models that show the potential impact of a given hazard upon the important elements of their society: people, properties, economies, and the natural environment.

The partners to this Field Guide are working to reduce the costs of collecting and curating data and transforming it into useful models that can guide policies and investments. Several have built non-traditional methods of creating exposure data, which tends to be the most costly to build and maintain.

---

Challenges: Constraints on the application of information to risk management

Most countries lack the resources, training, and software to place hazard and exposure data under a management process that allows for the assessment and mediation of risk. In many nations, the information necessary to catalyze this type of risk management thinking is blocked by a range of problems:

Fragmentation of Specialists

Risk assessment is a multidisciplinary process, but these experts rarely sit in one organization. Specialization has driven the design of modern bureaucracy towards hierarchies. While this structure is efficient for for transactions and coordination of workflows, the flow of information across (and between) organizations can be a challenge. Gatekeepers can prevent the timely flow of information, or may limit it in ways that hinder its use and reuse by others outside the original organization.

Data Fragmentation

Multidisciplinary analysis requires data from across specializations, yet these data are often segmented into silos. They may be in proprietary formats or locked under intellectual property licenses that require expensive payments. Some ministries may charge other parts of their own governments for use of the data, or might even have installed platforms that allow others to access the data but not download it.

Data Duplication

While donors may not set out to fund two or more collections of the same data, the result of having closed data is often just that. One ministry may not know what that other ministries possess or are currently collecting. This problem becomes more acute when NGOs are involved, as communication across partners is often not as good as it might be. Fusing separate datasets may not be possible, or may be very costly, if the groups are using different standards, software, and practices around its collection and quality assurance.

Data Access/Availability

While policies may allow data to be made available, the data curators might limit access or use of the data to specific parties. In this sense, access is discriminatory: it is only for certain approved entities.

Data Staleness and Incompleteness

Data may reflect best knowledge from an investment made more than decade before. In some countries, the last census or high-quality map may be decades old. Data about exposure may have never been collected at the level of resolution necessary to build risk models. Data may also be outdated and/or incomplete.

Exposure Data can be Expensive to Collect and Maintain

Data about buildings and built environment is resource-intensive to collect and maintain (time, cost, personnel, etc). The stock of buildings and infrastructure changes at the rate of construction minus the rate of destruction, mediated by a range of other factors, including the the rapid increase in population and the rate of urbanization. In mathematical terms, the stock of buildings is technically the partial integral of the rate of construction minus the rate of construction for a given time period between t(0) and t(n): ∫(construction_rate - destruction_rate).

---

Policy Challenge: underlying risks are accelerating

The processes of urbanization, population growth, accelerating rates of poor construction and urban planning, and increase in subsidence are changing the nature and magnitude of risks many developing countries. This is particularly pronounced in those countries most at risk of increased cyclones, floods, and droughts in cities with swelling peri-urban slums sited in the most vulnerable areas. For these policy makers, it is become ever more difficult to get a handle on dynamic risks.

That said, even dynamic risk can be mediated and managed.

The levers to manage risk are not always obvious. They may be at level of policy, such as building codes. They may require direct investments in retrofitting infrastructure to higher seismic standards. Or they may require soft infrastructure in a better prepared populace who stockpile supplies because they expect to be cut off from outside aid for a period of several weeks after the next major disaster.

Managing dynamic risks requires higher resolution data

Most governments function with data collected at best annually. In many places, data on the built environment have not been updated in decades, or are collected across a small sample of the country. Land cover estimates may be at greater than 1km or even 5km resolution (average building abstracted from sample across grid square.) There may be policy barriers to collecting the data, particularly where surveys of informal settlements by government officials might create political pressures to turn peri-urban slums into recognized municipalities. Yet, to assess the risk in places facing 5% annual population growth and increased probabilities of droughts, fires, floods, and landslides, governments need higher resolution data, both temporally (collection intervals) and spatially (grid square averages).

But the collection of required components for risk evaluation are no longer activities that most governments can afford to do alone; there is an increasing need for collective action.

To collect higher resolution data in times of economic uncertainty and tight budgets is a difficult choice. Professional surveys of urban areas can be very expensive. Analysis of the data can also be costly. But there is another way: collective action.

In Indonesia, Nepal, and a growing number of countries, governments have been mobilizing their ministries and citizens to collect and curate the data necessary to make everyone safer. Because much of the labor is done by community organization, the resulting maps of built environment are being created at ultra low cost. Working together, members of the private sector, public sector, and community organizations are building a shared understanding of their probable futures. With the aid of risk managers, the government is guiding a conversation about how to invest in resilient communities.

This need for collective action is emerging at time when communications tools and practices are introducing disruptive changes in the methods of coordinating collective action.

When communications are expensive, coordination generally occurs in centralized hierarchies, where decision makers at each level have specific authorities. Since the widespread use of the Internet, the costs of communications have been falling dramatically. In parallel, ever more computing power have been concentrated in devices whose production costs are falling according to Moore’s Law. (sidebar). As a result, billions of citizens of the world have access to low cost communications via handheld devices that allow for data collection, geolocation, and photography. They can organize their operations as swarms, collective intelligences that follow resilient networks of the Internet shaped into human form.

---

Open Data for Resilience: Harnessing Collective Action

Integrating these opportunities and new technologies has become the mission for a two-year old initiative called Open Data for Resilience. This initiative has drawn together international institutions, client governments, and community organizations harnesses collective efforts to create, collate, and analyze data about natural hazards and the risks that they poses to a nation.

These partners include a mix of donors, science agencies, and development institutions. Each has take a slightly different role in the construction of open data for resilience:

(note to each partner: this is a space for you to write a paragraph on your open data work)

Geoscience Australia

Global Facility for Disaster Reduction and Recovery

UNDP

UNICEF

UNISDR

United States Agency for International Development