1. Global Flood Model
December 2011

2. Endorsement
Dear Peter,
Thank you for the opportunity to meet with the 'volunteer' team. I enjoyed the opportunity not only to
learn more about the initiative but also to meet all of you who are making this joint effort work.
I have since we met in fact presented the idea to the senior management of FEMA USA. Clearly
interested in the concept as it includes what they do not (yet) have; the modelling of future flood
risks. I invited them to join the effort. And I think we may receive a positive answer.
I will in the coming weeks introduce it to others as well.
Yes, you can certainly use my ' name' when you present the idea and the concept - as supportive and
engaged in working to get it piloted and tested as soon as possible. I will meanwhile, also ask
colleagues here in the Secretariat to familiarize themselves with the concept and yes, be ready to
comment on the project proposal when you are ready to share it.
I look forward to our continued work on the Global Flood Model.
With best regards
Margareta Wahlstrom
Special Representative of the Secretary-General for Disaster Risk Reduction
United Nations International Strategy for Disaster Reduction (UNISDR)
Email: wahlstromm@un.org
Website: http://www.unisdr.org

3. The challenge to which the Global Flood Model responds:
 The Pakistan floods in 2010 (as one recent example) killed thousands, displaced
millions and materially reduced the country’s GDP for years to come.
 Australia, Brazil, Philippines, Poland, Thailand, Turkey and USA and others have
also experienced severe-to-catastrophic flood impacts since the Pakistan event.
“For any given flood-prone location, how can we create a global community
platform that enables better understanding of the complex causes behind the
risk of flood impact - the better to support the decisions, and also the
education and communication, needed to mitigate that risk before, during
and after the flood event itself?”
 “Flood impact” = any adverse economic, social or
environmental consequence
 Covers hydrometeorological or coastal flooding

4. 1. What will the GFM do?
The GFM will provide, for any flood-prone location on earth, tools for:
– Predicting the risk of floods, and for devising long-run mitigation strategies such
as land-use changes and infrastructure improvements.
– Forecasting the timing, location and severity of specific flood events, and
enabling effective emergency response and post-flood recovery.

5. The GFM covers the entire causal web behind flood impacts
See appendix for coastal inundation variant of this chart

6. The GFM enables a holistic, integrated response to flood risk
Hazard
Vulnerability
Risk characterization, Exposure
- Failure modes and
based on: - Geographic extent of
resilience of critical
- Climate trends hazard
infrastructures
- Rainfall ground - Location of economic
- Effectiveness of flood
interactions activity, populations,
defenses
- River and lake critical infrastructure
- Building codes
behavior
Mitigation & Recovery Resilience & Adaptation
- Engineered “flood - Knowledge of risk and
proofing” mitigation
- Land use decisions, siting - Natural systems –
of critical assets marshes, mangroves etc
- Relocation - Short and medium range
- Emergency planning forecasting of floods and
- Costs and benefits impacts

7. Why now? Because understanding, data and IT are catching up with need
 For the first time in history, the causes
of flood impacts can be understood and
addressed in a holistic way:
 Flood impacts such as that in
Pakistan seem to be – Scientific and engineering
increasing, driven by: understanding of causes and
mitigation strategies are growing;
– the impact of climate
change on rainfall events; – Remote and in situ-sensed data
sets enable granular understanding
– The impact of of global and location-specific risk;
urbanization and poor
land-use decisions in – Modeling, optimization and
placing more people and visualization tools provide the ability
value in harm’s way; to use that data to make better risk-
management decisions;
– Sub-optimal flood event
response. – Processing power enables us to
process massive data sets, as well
as complex, integrated models and
frameworks, cost effectively.

8. 2. Who will benefit?
The GFM combines humanitarian and commercial motives. Its beneficiaries will be:
– The public, seeking to preserve personal safety and property;
– State and local governments, seeking to safeguard economic activity, and
improve resilience;
– NGOs, similarly seeking to respond proactively to flood events;
– The insurance sector, seeking to understand and price flood risk;
– Large corporations, seeking to protect global operations and supply chains.

9. Humanitarian and financial/economic benefits* from the GFM
*To be validated
Humanitarian and quantified
Loss of Life
 Lives saved
 Families saved (reduced
loss of parents, bread- Economic security
Social/Political Cohesion
winners)  Resilience:
 Shared understanding
 Protection of education,
 Property protection
infrastructure, social
 Poverty avoidance, quality
programs
of life protection
 Protection of economic
 Avoidance of population
activity (w/w benefit to
dislocation
Global Flood global supply chains)
Model
Health
Capital Markets
 Injury avoidance
 Reduction in risk –
 Water-borne disease
reduced cost of capital,
avoidance Economic Expansion increased health of
 Shelter and nutrition  New insurance markets insurance sector
protected  New service opportunities –
engineering, consulting
Financial & Economic

10. The GFM will have many groups of users, benefiting in specific ways
 Citizens: how can I understand the risks from floods to myself, my family and property and
either remove those risks or make effective plans for dealing with them?
 City or state managers, NGOs: how can I understand long term flood risk arising from the
interactions of weather, sea levels, hydrology and land use patterns?
– How can I make my city’s infrastructure and society more resilient?
– How do I forecast flood events and impacts, evacuate people, and help my city recover
after an event?
 Insurers: how do I understand, price and aggregate flood risk, so enabling me to offer cover?
 Corporate entities:
– How do I plan for flood risk to my operations, and/or my supply chain?
– How can I manage my flood insurance costs?
– How do I help the communities in which we operate become more resilient?
 Scientists or engineers: how can I advance the science, and contribute my understanding,
of the issues involved to help create practical solutions to these needs?

11. 3. How will the GFM be built?
The GFM is a set or “stack” of integrated modules, each composed of models and
data.
– There are two core elements: a framework of specifications for each module,
and a live “reference version” (a worked example) of each.
– Users may work with this reference version, or substitute their own models and
data. If these meet the specification for the relevant module, they should
interoperate with the rest of the GFM.
– Our intent is to improve local adoption and stimulate the development of multiple
– but compatible – alternatives, so strengthening mankind’s ability to manage
flood impacts.

12. Illustrative “stack” of modules in the GFM
 Set of integrated modules and Fluvial/pluvial “stack” Illu
str
indices, each containing ativ
e
models and data.
 Each module captures part of
the flood impact causal web,
and adds value in its own right.
 Each may require inputs (data,
calculations, indices) from
logically antecedent modules,
and provide outputs to others.
 Collectively, modules form the
complete “stack” covering the
causal web.
 Modules enable dis-
aggregated construction and
See appendix for coastal inundation variant of this chart

13. GFM specifications
 Role: what need does the module meet?
 Specifications for each module  Function: what does the module provide?
will:  Required outputs: content and semantics,
scale, frequency, tolerances/uncertainty levels
– Define the needs that the (inherent in each module and from
module meets, and how combination with other modules), formats,
– Define interfaces and ensure standards, screen layouts.
interoperability of the complete  Required inputs: as above, plus validated data
GFM “stack” sources.
– Enable local models and data  Models, functionality and data transformations
that comply with specifications required to generate the specified outputs.
to interoperate with the rest of  Application programming interfaces (APIs)
the GFM that allow interfaces to other modules and
also insertion of local models and content.
 Specifications will include a
means of tracking accumulated  Intended evolution: levels of function and
uncertainty through the GFM resolution that will be added in the future as
science and data availability permit.
“stack”.
 Technical issues foreseen.

14. GFM live “reference version”
 Prove the functionality of each module and
the GFM as a whole
 Provide “worked examples” to enable user
needs to coalesce around an actual system
as the GFM evolves  Reference versions will be
 Benchmark module/model skill and usability piloted for different
locations around the world
 Ensure global coverage in lieu of local
offering specific
alternatives (see later)
combinations of physical
 Provide reference points and examples to characteristics and risk
support development of local alternatives exposures
 Enable some models that are inherently – This potentially
global: exposure models, financial market provides a means to
models … engage key countries
 Enable end-to-end simulation and exploration
of “what-ifs” across GFM modules
 Support outreach and marketing activities

15. Global foundation, local implementation
 “Localization”:
– Local needs
– Local data, models, Country/ Country/ Country/ Country/
adaptations thereof location- location- location- location-
– Local expertise specific specific specific specific
– Implementation model model model model
and data- and data- and data- and data-
– Visualization
set set set set
– Feedback 1 2 3 n…
– Engagement and
commitment
 Specifications,
standards, reference
Global Foundation
version(s)

16. Evolution path – guided pragmatism
 Iterations of GFM modules will be prioritized by
value: where need exists, and where data,
model skill, resolution, usability and
accumulated uncertainty can be improved at
justified cost.
 The precise evolution
– The baseline (“Version 0.1”): will consist of path will be determined
modules that drive most variance in flood in a feasibility study
impact, and do most to improve risk that determines the
management and event response. best balance user
– Some modules may provide higher-level need, value, and
indices that enable deferral of others until scientific and
scientific advance or data availability makes computing possibility
greater precision worth the cost.
– Some modules may initially be applied
globally, while others will be local. We will
achieve greater granularity over time, again

17. 4. How will the GFM be organized and financed?
The GFM will be developed and managed by non-profit public-private foundation
created for the purpose.
– The business model will be derived from open source software (eg Linux): the
core specifications and reference version of the GFM will be licensed free for
non-profit usage.
– For profit users such as software companies, engineering companies and
business or risk management consultancies will pay an annual license fee. This
will contribute to the upkeep and maintenance of the GFM.

18. Key principles
 The GFM will be created and managed as a public good, by a non-profit public-
private foundation established for the purpose.
 Its core IP will be open-sourced:
– It will be placed in the public domain and given appropriate legal protections to
enable unencumbered use.
– Access will be free of charge for non-profit use
 Following the model of open-sourced initiatives such as Linux, we will stimulate an
“ecosystem” of for-profit users –engineers, risk consultancies, software companies
and others - who add value to the core models for their own commercial purposes.
– Fees that ecosystem users pay will cover GFM’s running costs and perhaps
make it self-funding in the longer run.

19. Business model
 Founder members -
core group who set
direction and oversee
design of GFM
 Affiliates - make a
smaller donation than
founders in exchange for
early access to
deliverables, or (in case
of cities) the chance to
become a pilot location.
 Founders and affiliates
include insurers/re-
insurers, large
corporates, national
governments, and
supranationals (UN,
World Bank and others)
 Ecosystem (“Accredited users”) - ecosystem members would pay certification/ accreditation
and license fees for use of core IP – idea is for GFM to become self-funding over time.

20. Revenue structure
 Launch funding:
– Donated by founder members (would receive some level of free access to all
deliverables over some period of time to be determined)
– Affiliates will make smaller donations in exchange for chance to become a pilot
location, or early access to deliverables.
– Wealthier countries may be also invited to purchase blanket access for non-
profit users within their boundaries).
– Required donation sizes to be determined. We estimate that we will require ~
$50 million, to cover a 5 year development program, to be confirmed in
feasibility study. Work would begin well before this sum is raised
 Ecosystem – illustratively, world wide:
– 50 for-profit users, paying $25,000 per year and 200 for-profit users paying
$10,000 per year = $3.25 million per year to support ongoing running costs.

21. Organizational Model – based on Global Earthquake Model (GEM)?
 As an analogous program,
GEM has solved many of the
problems that GFM will have
to address.
 It may therefore be a model
for solutions to the
organization and
management issues that
GFM will have to address:
– Attracting initial funding
– Enabling peer reviewed
science, in the service of
carefully defined user
needs
Modeled after Global Earthquake Model
– Enabling cross-disciplinary input – several branches of science, engineering
– Enabling national needs and aspirations to be heard and accommodated
– Execution, progress tracking and issue resolution within a complex, multi-year program
– Transparency and accountability
 Some data files that GEM has assembled – for example
– Outreach building or infrastructure inventories in specific areas – may
in principle be reusable for GFM.

22. 5. Next Steps
The GFM is being created by Willis Reinsurance, UK Met Office, Esri, Deltares and
IBM, who are each committing substantial pro-bono resources to the initiative. Our
first task will be to create a demonstrator, which will be available early in 2012.
– We will then require donor funding in order fully to investigate the feasibility of
the concept and commence the process of building the GFM.

23. Demonstrator and Beyond
 We will undertake a feasibility study, that in
 We are currently creating a the light of reactions to the demonstrator:
demonstrator/proof of – Defines the overall program for the GFM,
concept for the GFM, which including completion of the baseline;
we hope to have available by
Jan 2012. – Captures initial user requirements;
– This will provide a focus – Proposes a development sequence and
for attracting views and evolution plan;
inputs, and for fund-raising – Identifies existing and planned models
efforts while also providing and datasets that we can use;
measurable value in its
own right. – Identifies the required organization
structure, business and IP model;
– The demonstrator will
become a large part of the – Defines the required technology
“baseline” GFM referred to architecture;
earlier – Creates detailed plans and business
case.

24. Appendix: Additional detail

25. Complex webs of causation – coastal inundation

26. Coastal inundation “stack”

27. Localization: “module by location*” structure
*The definition of
“location” applies at
multiple scales –
grid square,
suburb, city,
country, river basin,
region - depending
on the nature of the
phenomenon, the
ability of the
science to add
value and the focus
of the decisions
being made.

28. Global foundation: supporting elements for the GFM (1/4)
 Requirements set
– Who are the users and what do
they want? What are the “use
cases”?
– What do users need the GFM to
do, in what timescales, at what
scale and accuracy? What
services do they require?
– What is the value?
 Systems dynamics model
– Describes causal chains behind
flood impacts of different kinds
– Captures the “reference set” of
causes we will attempt to model
– Helps define boundary of each
module and required
interactions

29. Global foundation: supporting elements for the GFM (2/4)
 Reference data model
– Definitions, frequencies,
tolerances, standards of key
data items exchanged
between modules
– Integrated into a single
ontology and semantic model
for all GFM elements
– May or may not be based on
existing standards
 Reference model library
– Collection of models and tools
adopted in GFM modules, plus
interfacing & co-calibration
needs
– Supporting documentation –
how to use each model
appropriately

30. Global foundation: supporting elements for the GFM (3/4)
 Reference dataset library
– Collection of global/local
datasets created or adopted in
GFM modules
– Supporting documentation –
how to use each dataset
appropriately
 Quality control framework:
– Specifies uncertainty levels,
tolerances, for each module and
the stack as a whole
– Identifies valid combinations of
modules, datasets and models,
by version
– Specifies validation and peer
review process for each module,
and records outcomes
– Service levels

31. Global foundation: supporting elements for the GFM (4/4)
 Computing platform
– Supports delivery of GFM
modeling service
– Technology architecture, including
hardware, operating software and
core tools (eg database, GIS,
simulation engine and etc)
– Collaboration toolset –
exchanging model runs,
comparing results etc
– Driven by requirements to run
models etc - reference standards
as well as physical products
– Probably “cloud” based, to enable
access and scalability to a
worldwide community
 May become separate business in its
own right?

32. GFM and scientific advance
 The GFM requires the active
engagement of the scientific
community in many domains.
 It will provide the framework for, and
catalyze, scientific and engineering
advances required for its evolution.
 As a major example, one area where
advance is required is hydro-
meteorology and the creation of a
“global rainfall model” – the GFM
modules within the red box, right.
 Potentially, other required areas of
advance are:
– Management of urban run-off
– Financial market modeling

33. Development methods
Method Advantages
 Multiple smaller iterations  Demonstrates progress
rather than “big bang” changes  Smaller funding amounts can be tied to specific increments
 Reduces technical risk – if any one increment fails, less impact
 Specifications  Creates envelope in which development to each module need to “fit”
 Enables disaggregation of workload to multiple parties
 Enables local inputs and local creation of models, data etc
 Rationale for, and expectations of, each module are clear at all times
 Prototyping (reference  Engages users (and creators) visually
versions) – design GFM  Helps ensure fit with user requirements
modules “from the user screen
backwards” and early releases  Provides means to engage with potential sponsors
 Multi-disciplinary teaming  Essential in project such as GFM where multiple inputs needed to
(within each module and over complete the whole – regular interactions maximize shared
the whole) understanding of the whole
 Open development methods  Engages community and ensures “ownership”
(users as co-developers, early  Proven capability in rapid, accurate development of complex integrated
releases, modularity, etc) software (Linux etc)
 Peer review of models and  Ensures scientific rigor and validity
specifications