Progress in Disaster Science 12 (2021) 100203

Contents lists available at ScienceDirect

Progress in Disaster Science

journal homepage: www.elsevier.com/locate/pdisas

Creating resilient communities with medium-range hazard warning systems

⁎
Bapon (SHM) Fakhruddin a, , Peter Gluckman b, Anne Bardsley b, Georgina Griffiths c,
Andrew McElroy d
a
Tonkin + Taylor, New Zealand
b
Koi Tū: The Centre for Informed Futures, University of Auckland, New Zealand
c
MetService, New Zealand
d
United Nations Office for Disaster Risk Reduction, Fiji

A R T I C L E I N F O A B S T R A C T

Article history: Resilience to natural hazards requires integrated risk management – from hazard identification and risk register, haz-
Received 14 July 2021 ard warnings and risk communication, to preparation for and response when an event occurs. Medium-range hazard
Received in revised form 18 August 2021 warnings play a vital role in strategic and tactical planning for resilience to natural hazards. Research advancements
Accepted 29 August 2021
have rapidly improved observation, modelling and analysis of natural hazard forecasting for disaster risk reduction.
Available online 20 September 2021
However, many of these advances have not been tailored to benefit communities at risk as geophysical and hydrome-
Keywords:
teorological hazards continue to claim thousands of lives and cause irreparable damage to homes, businesses and crit-
Medium-range hazard warning systems ical infrastructure. This paper discusses the opportunities to use medium-range ensemble forecasting to support
Ensemble forecasting decision making to increase the resilience of communities and the ways to embed holistic early warning systems in
Subseasonal-to-seasonal numerical weather risk-informed sustainable development. It provides a set of recommendations for medium range forecasting applica-
prediction tions for people-centred early warnings and disaster preparedness.
Risk register
Risk communication
People-centred warning

1. Background responsibilities, thus ensuring good risk governance [40]. Ideally, risk reg-
isters highlight and assist in systemic risk and cascading and compounding
With the current trajectory of global climate change, countries face a effects [7,11,16,54] and are linked to ensure an effective people-centred
continual increase in the frequency and intensity of extreme weather early warning systems. Outlining risks in a public risk register enables effec-
[52]. This has gradually led to a greater emphasis on climate and disaster tive risk communication, which is key for any early warning system
risk assessment, and identification of adaptation, mitigation and disaster [58,59]. Particularly, dissemination/notification and community capacity
risk reduction options. Early warnings are crucial in disaster risk reduction, (connection and response) are two interconnected fundamental elements
as they empower communities to prepare for natural hazards [58] and, if within an early warning system to better understand local context, culture
effective, can prevent loss of life, protect property and livelihoods, and re- and sensitivity [23].
duce the economic impacts of disasters [72]. The United Nations Global As- With the recent enhancement of numerical weather forecasts and meet-
sessment Report [79] revealed that investing USD 6 billion in appropriate ing the demand of increase lead time, there is a shift from single determin-
disaster risk management strategies can generate benefits of USD 360 bil- istic weather model outputs to probabilistic weather forecast information,
lion. Similarly, it has been demonstrated that one dollar invested in disaster generated from ensemble-based prediction systems. The concept of using
preparedness and reducing vulnerability of people can prevent six dollars' an ensemble (group) of weather models to obtain a range of possible future
worth of disaster-related economic losses [22,26]. weather outcomes partially side-steps the issue of initialisation error and
A risk register plays a critical role in disaster risk reduction by identify- weather model error and allows predictions across longer time periods. En-
ing and documenting risks, their likelihood and impact on the system of in- sembles are now commonly used operationally to predict weather out-
terest (e.g., a community, country or region, a business or sector). Risk is comes over the 7-to-30-day validity period and potential inputs for the
typically placed in a register in the form of matrix consisting of risk scores hazard modelling such as flood, droughts, cyclones, etc. [10,42]. The eval-
(using hazard, exposure and vulnerability assessment) [38,48]. A risk regis- uation and improvement of forecasting and warning system been changed
ter outlines proactive actions to mitigate the risks with assigned time to meeting lifesaving needs, meeting people's demands, leveraging

⁎ Correponding author.
E-mail address: [email protected] (B.(S.H.M.) Fakhruddin).

http://dx.doi.org/10.1016/j.pdisas.2021.100203
2590-0617/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/
4.0/).
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

Fig. 1. Evaluation of ensemble-based forecasting systems since 2000 (Source: Authors).

resources and economic benefits of sectors. Fig. 1 presents a brief summary 2. Political challenges for effective end-to-end early warning systems
of evaluation of ensemble-based forecasting systems over the last two de-
cades. Risk registers can highlight rare but forecastable events that experts can
The initiatives of the subseasonal-to-seasonal (S2S) timeframe focus on describe in terms of likelihood and impact. What cannot be identified with
period-averaged forecasts, usually weekly, at lead times of weeks to months any certainty is when the events will occur. Because some events are ex-
[29]. The S2S timeframe is considered difficult to forecast, since it is largely tremely rare, but their impacts are high, the challenge is how to communi-
beyond the influence of initial weather conditions but not long enough for cate to governments, organisations and communities of interest to invest in
significant variation in slowly evolving earth climate system components resilience preparation and risk mitigation, including in early warning
(such as the ocean, soil moisture and sea ice) to add predictive skill relative systems.
to persistence [66]. However, there is growing demand for medium- The issues that lead the public sector to not prepare for risks associated
range forecasts (7–30-day validity period) from a number of sectors (e.g. with hazards are generally not technical; rather they largely relate to risk
agriculture) [24]. perceptions and cognitive biases of the decision makers [31]. Communicat-
There are numerous opportunities provided by ensemble-based pre- ing risk and actions to the policy and political community is never a simple
dictions to enhance medium range forecasts. For instance, “Anticipation technological discussion, it is a complex matter of dealing with cognitive
Hub”, a platform hosted by German Red Cross along with International biases and political realities [76].
Federation of Red Cross and Red Crescent Societies (IFRC) to facilitate A major bias affecting most people, including politicians and policy
exchange of knowledge, “learning, guidance, and advocacy around antic- makers, is myopia or short-termism: a focus on immediate needs at the ex-
ipatory action both virtually and in-person”, uses forecasts or warnings pense of longer-term planning – and for the politicians, the length of an
to carry out early actions and finance anticipatory actions (Anticipation election cycle is critical [31,34]. Prevention of disasters that may occur in
[5]). Forecast-based Financing (FbF), a programme developed under the the future gets no political credit if an event does not occur during the pol-
Anticipation Hub, predicts disasters to prevent their impacts using de- iticians' mandate. There is always an opportunity to procrastinate preparing
tailed forecast information and risk analysis. As of 2020, FbF projects for major events until they happen, by which time it is too late. There is also
are implemented across Africa, America and Asia-Pacific ([37]; Anticipa- a tendency to forget lessons of the past disasters [9].
tion [5]). The World Meteorological Organization (WMO) initiatives on Policy makers must first recognise potential risks in order to establish a
multi-hazard impact-based early warning system and Climate Risk and system that scans for change in identified risk parameters and alerts to a po-
Early Warning System (CREWS) are applied in many countries to tential event. In a traditional, idealised view of early warning system, the
achieve the global target G (early warning system) in the Sendai Frame- warning system acts as a “searchlight” on society - it finds a problem, alerts
work for Disaster Risk Reduction and enhancing lead-time forecasts the government and the problem gets addressed. The reality of this is more
[14,23]. precarious and the stability of society can often balance on the accuracy of

2
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

early warning systems, leaving little room beyond survival in the face of a forecast is useful to a recipient only if the weather variables it predicts are
major event or disaster. Lessons from historical events are motivation for relevant and the forecast is timely for the recipient to make an action. Use-
governments to improve early warning systems; governments need to ful forecasts are those that meet recipients' needs in terms of such attributes
know what lies ahead to help maintain civil and political stability. as timing, lead time, and currency; climate parameters; spatial and tempo-
An imbalance exists between the user expectations and what fore- ral resolution; and accuracy. Because the usefulness of forecasts is depen-
casters, or national meteorological services, can provide to the community. dent on both their accuracy and their relationship to recipients'
As global climate changes, we need to use technology to enable various informational needs and coping strategies, the utility of forecasts can be in-
scale forecasting models to predict weather as accurately as possible and creased by systematic efforts to bring scientific outputs and users' needs to-
bring them to operational use. Science can provide informed services up gether.
to a certain limit, but user expectations always demand more. When expec-
tations are not met, users are less willing to use the information and take 3.1. Weather, water and climate service framework
action. It is necessary to find a balance between the user expectations and
the services that forecasters can provide. Managing expectations and A framework for weather, water and climate services is a policy and in-
explaining fully the limitations of forecasts would enable users to under- stitutional mechanism to strengthen the capacity and streamline the sys-
stand why they may not be able to receive all the information they want, tems for weather, water and climate services for enhanced decision
and why information may be uncertain. As climate change intensifies future making. The WMO [70] developed guidelines for strategic plan and frame-
droughts and floods, the importance of this community-based anticipatory work for weather, water and climate service to enhance national meteoro-
approach will grow. logical and hydrological services' capacity [12,69–74]. Several countries
already adopted a national strategic plan and framework for weather,
3. Medium-range ensemble forecasting applications water and climate services. The potential increase in occurrence and inten-
sity of extreme weather events as a result of climate change, coupled with
Ensemble-based prediction is now commonly used at most national increasing population in vulnerable areas, reinforces the need for seamless
weather services [10]. An ensemble of models is run with varying amounts integration of weather, water and climate information systems (Fig. 3).
of error associated with weather conditions at initialisation for an extended
period of time, which gives a probabilistic range of possible future weather 3.2. Day-to-day forecasting: adopting ensemble forecasting for dealing with un-
states. This gives decision makers viable statistical information over ex- certainty
tended validity periods (7–30 days in most cases).
Common uses of ensemble forecasting include tropical cyclone track Errors in weather forecasting can be attributed to two causes: inaccurate
outcomes. For example, 15% of ensemble members may track a tropical cy- initial conditions and deficiencies in the weather model [8]. Analysis error
clone formed in the south-west Pacific towards northern New Zealand, —having an incorrect or incomplete representation of the atmospheric
while 85% of ensemble members show that it will stay in the tropics. This state at initialisation —is one source of uncertainty. Given the infinitesimal
split in outcomes is useful information to decision makers, as percentiles scale on which the world and atmosphere operate, it is a practical impossi-
can be worked out to assess risk; this would not be possible if just a single bility to devise a perfect representation of the weather system state.
weather model was used. Another common application of medium-range Model error also has a strong impact on the evolution of forecast skill
ensemble forecasting are precipitation probabilities. Forecast ranges and [15,53]. When many different types of forecast models are used to generate
probabilities of precipitation totals over an extended duration can be useful an ensemble forecast, the approach is termed multi-model ensemble forecast-
information for hydro-storage lake level managers who want to know how ing. This method can improve forecasts when compared to a single model-
full their lake may be and manage lake levels accordingly. Generally, the based ensemble forecasting. Forecasts are only as good as the language and
ensemble mean (50th percentile) output is the most commonly used statis- method used in communicating to affected communities. Ensemble fore-
tic from an ensemble forecast; however, knowing a range of probabilities al- casting is a good approach to gather a range of forecast weather outcomes
lows decision makers to assess what the width of the possible forecast and to produce probabilistic information from which agencies can assess
outcome is (Fig. 2). risk and inform decision making. It is a good way to deal with both
Medium-range forecasts are inherently uncertain due to the chaos in the initialisation error and model error uncertainty for the medium-range (1-
atmospheric system; moreover, forecasting skill varies geographically, tem- to-14-day horizon in particular). However, because ensemble forecast in-
porally, and by climate parameter [66]. A medium-range to long-range formation is significantly different from the usual ‘deterministic’ weather

Fig. 2. Ensemble forecasting concepts (Source: MetService, 2021). This diagram shows initial condition uncertainty at analysis time, with multiple models producing differing
forecast solutions through time (depicted as the forecast uncertainty). The forecast outcomes from individual ensemble members can differ considerably (selected model
outcomes are depicted by white crosses), but we do not know which outcome is more reliable [33].

3
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

Fig. 3. Value chain of risk-informed sectoral development of weather, water and climate services (Based on [71])

Fig. 4. Forecasts skills and integrations (Source: [78]).

4
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

predictions seen on TV news, careful presentation of weather outcome ver- 3.4. Improving forecast of high-impact weather events
sus predicted chance of occurrence is essential. To enhance community re-
silience and disaster preparedness, there is a need to involve communities High-impact weather events pose threats to human life, property and
at risk, facilitate public education and awareness of risk and potential im- the economy, and inflict significant societal hazards. It is, therefore, crucial
pacts, and disseminate warnings in a timely manner [70]. to have skilful forecasts of the risks with sufficient lead time to make appro-
priate precautions. As climate change causes global temperatures to rise
and more extreme weather events to occur, researchers need to continue
3.3. Integration of short-, medium- and long-term and subseasonal forecasting improving forecasting and modelling to be as accurate and as far in advance
applications for disaster response as possible [27]. Effective forecasting of high-impact events often requires
perspectives and tools that are different from routine forecasts. A number
Many national hydrological and meteorological services are moving to- of factors make advance forecasts of high-impact weather events challeng-
wards a multi-hazard impact-based forecast and warning services that ing, including insufficient resolution to simulate hazard scenarios dynami-
translate hydrometeorological hazards into sector and location specific im- cally in a forecast model; model biases in representing storms which often
pacts and responses [71]. Effective impact-based forecasting requires col- become increasingly pronounced in extreme scenarios; and even difficulty
laboration with other decision makers including disaster and emergency in defining and verifying the high-impact event. New technologies
managers, stakeholders and communities [74]. Each day of additional (e.g., machine learning and artificial intelligence) and international collab-
warning provides emergency managers the time to inform communities oration may accelerate these efforts.
at risk to prepare for extreme weather and climate related shocks (Fig. 4).
Short-term models, like those readily available on smart phones, use 3.5. Risk register for proper risk assessment
current weather observations to make predictions. For forecasting fur-
ther into the future, the weather at time of initialisation becomes less Risk register can inform the authorities on the most significant risks that
relevant. Long-term seasonal outlooks predict how different the climate could occur in future that might have huge impact on the nation or commu-
will be compared to normal over the next three months (i.e., will it be nity. It serves as a guide to identify the areas that need actions in a nation or
hotter or colder). This high-level perspective on how regional climate community [13,67]. Risk registers facilitate a logical and efficient approach
is expected to vary is based on the slowly changing planetary patterns towards the risk management. The National Risk Register developed by the
which drive the weather over a number of months, such as intermittent Government of United Kingdom presents an assessment of the likelihood
oceanic warming patterns like El Niño and the extent of sea ice coverage and potential impacts of risks related to environmental, human and animal
in the Arctic Ocean [69]. health, major accidents, societal, malicious attacks, and risks occurring
Subseasonal forecasting lies between short-term forecasts (up to overseas that might affect the United Kingdom [32]. For instance, fatalities,
10 days) and seasonal forecast (one month to several months) [69]. Mak- damage to property and infrastructure, and disruption to essential services
ing such forecasts is difficult because the initial information that drives were some of the identified risks posed by earthquakes, and regulations,
the short-term models is no longer as useful, and the long-term climate seismic hazard studies, models and forecasts, seismic risk mapping, and re-
drivers associated with seasonal forecasts have not yet become apparent search projects on the United Kingdom earthquake activity were a few mit-
[69]. Subseasonal forecasting has been ignored in the past, due to how igation measures to the earthquake risk [32].
difficult it can be. It is improving slowly but steadily, largely due to ad- The reality is complex and system-based thinking is crucial [6]. An early
vanced technology to produce better computer models and new insights warning system factoring in compounding and cascading hazards and a
about the atmospheric and oceanic patterns that drive weather over the seamless integration of short-range, medium-range and long-range weather
long term [68]. and climate forecasts could enhance the decision-making process. With a
Many weather patterns, some of which are not predictable, influence change in our mindset using advanced seasonal warnings, we can produce
the weather on the subseasonal scale. Some weather patterns, such as the more useful, usable information to help industries, sectors and communities
Madden-Julian Oscillation (MJO),1 have been researched to improve haz- to better understand foreseeable risks, scenario analysis and adaptation op-
ard predictions [69], but more advancements in modelling the MJO events tions.
are still needed. Another weather pattern which needs to be understood to
improve subseasonal forecasting is a “sudden stratospheric warming”
above the Arctic or Antarctic. It happens every couple of years in the north- 3.6. Dealing with uncertainty
ern hemisphere and less often in the southern hemisphere and can affect the
weather worldwide.2 Periods in which these events occur is called “fore- Risk communication is an essential component in reducing disaster risk,
casts of opportunity” as during this time subseasonal forecasts become especially for medium-range hazard warning systems [19,23,80]. Effective
more skilful. An analysis conducted by Giuseppe et al. [30] revealed that communication through hazard forecasting is critical for communities to
the skills provided by the ensemble prediction systems extend beyond understand risks and take appropriate actions. As we face a complex envi-
10 days when compared to the deterministic models. ronment where risk is more systemic, national, and local governments
Ensemble forecasts can be used to generate flood thresholds as floods need to have ability to successfully communicate hazard risk and actions
are the most frequent disaster event that occurs across the world [18]. through a coordinated all-of-society approach (Fig. 5).
Thresholds that reflect the flood event frequencies of the real-time ensem-
ble forecasts across all forecast lead times is crucial for accurate forecasts 3.7. Impact-based forecasting
[77]. According to the study on “Ensemble reforecasts to generate flood
thresholds for improved global flood forecasting” (by [77]), reanalysis- Impact-based forecasting allows practitioners to interpret the hazard
based thresholds were found effective for the first one to four days of the warning and requires strong partnerships between hydrometeorological
30 days forecast range. services at national level and sectors functioning in disaster risk reduction
and management [72]. It results in people taking early actions such as evac-
uating vulnerable communities, individuals, and their livestock, “pre de-
ployment of flood barriers”, and closing roads and bridges [73]. Aspects
1
The MJO is an eastward moving tropical disturbance of clouds, rainfall, winds, and pres- of the Sendai Framework for Disaster Risk Reduction, the Sustainable De-
sure that traverses the planet every 30 to 60 days on average.
2
For instance, a northern stratosphere warming often causes storms in the United States. A
velopment Agenda, and the aims of the Risk informed Early Action Partner-
large southern stratosphere warming set up the weather events which led to the tinder-dry con- ship and the Anticipation Hub could be achieved by impact-based
ditions in Australia and caused major bushfires in late 2019 and early 2020 [45]. forecasting [5,72].

5
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

Fig. 5. User needs and application for forecasts (Source: [23])

When developing and implementing an early warning system, we need forest fires, earthquakes, droughts, and volcanic eruptions [21]. Countries
to consider that a more widespread application of early warning systems in these regions have limited to no infrastructure for early warning systems
equates to a more widely experienced uncertainty. Even when there is and are inadequately prepared and extremely affected by these disasters
warning information available, especially for sector and strategy decision [28]. Pacific Island Countries (PICs) are far behind the implementation
making, it does not immediately translate into successful application, as and applications of medium range ensemble forecasting system. Currently,
there is a time horizon to turn strategic decisions into actions. all Pacific Island countries are providing basic level of climate services that
Impact-based forecasting utilises either deterministic forecast data or include climate outlooks, summary of statistics, ENSO updates, and climate
ensembled forecast data for generating results. It is proven that ensemble- change information [28]. “Explore the use of ensemble techniques in trop-
based forecasting data produce more consistent and successive forecasts ical cyclone forecasting and probable forecasts” is one of the key priorities
than other systems [56], hence, it should be used for determining the of Pacific Islands Meteorological Strategy 2012–2021 [60]. The ensemble-
impact-based forecasts. based forecast warning systems could benefit PICs and SIDS in preparing
for adverse weather conditions, prepare to response to hazards and mini-
3.8. Socio-psychological aspects of the end users mize the risks and enable faster responses.

Ensemble forecasting allows users to have reliable forecasting of the 3.10. Resilience through medium range forecasting
range of most likely future scenarios. It provides estimates of the atmo-
sphere and assists in generating future climate projections [10]. Schroeter The ensemble forecast models presents the practitioners and users with
et al. [56] reported that “thresholds on probabilistic forecasts derived most likely scenarios of the weather and supports them in being prepared to
from EPS can be set to alert forecasters and other users to the likelihood face the disasters and assists in reducing losses and faster recovery [49]. For
of a hazard occurring, which can substantially increase the lead time of use- instance, New Zealand is susceptible to natural and climate hazards such as
ful forecasts for end users”. earthquakes, volcanic eruption, tsunami, floods, cyclones, coastal erosion,
Besides, a wide range of psycho-social benefits could be gained with the sea level rise and extreme weather events. New Zealand's National Disaster
assistance of ensemble-based forecasting. For example, the value to Resilience Strategy focuses on creation of resilient communities by manag-
weather forecast to road transport carried out by NZIER [50] revealed ing risks, effective response to recovery from emergencies, and enabling
that direct costs such as loss of life and life quality, vehicle damage, medical empowering and supporting community resilience. Strategies were devel-
expenses, legal and court costs, and loss of productivity could be minimised oped across social, cultural, economic, built environment, governance,
with the availability of accurate forecasts. The 1–10 days flood forecasting and natural sectors. One of the priorities is to improve the availability and
system has proven enormous benefits to the community in Bangladesh accessibility of multi-hazard early warning systems in New Zealand [47].
[24,35]. The European flood alert system (EFAS) is widely used for plan- As of 2021, the country is underutilising the medium-range forecast ser-
ning and operational uses. Additionally, forecasts may reduce or avoid vices, however, with the adequate infrastructure facilities these forecast
some intangible costs such as trauma, mental illness, bereavement, loss of services could be enhanced [49].
trust in authorities, and loss of jobs (societal disruptions) [50].
4. Conclusion
3.9. Benefits of ensemble-based warning system for climate-sensitive countries
Issuance of reliable forecasts and warnings in a form that is readily un-
Small Island Developing States (SIDS) and the Asia region are one of the derstood and creating awareness on how to prepare against such hazards
world's most disaster-prone in terms of the recurrence and severity of natu- before they become a disaster is necessary for any warning system [72].
ral hazards, with high exposure to floods, landslides, tsunamis, cyclones, With the advances in science and multi-model ensembles, warning skills

6
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

have increased and have the potential to reduce disaster risk and sectoral [22] Fakhruddin BS, Schick L. Benefits of economic assessment of cyclone early warning sys-
tems - a case study on cyclone Evan in Samoa; 2019 [Progress in Disaster Science].
decision making as across fields such as water resource, agriculture, [23] Fakhruddin Bapon, Lomarda Nanette, Boylan Kate. Total warning system for tropical cy-
dairy, livestock, energy, epidemics and ecology. Weather never stops, it is clone. PreventionWeb, UNDRR; 2019https://www.preventionweb.net/publication/
ever-evolving, and is never quite the same twice. It affects almost every as- total-warning-system-tropical-cyclones.
[24] Fakhruddin SHM, Kawasaki Akiyuki, Babel Mukand S. Community responses to flood
pect of society in some capacity, from daily life to the economy. However, it early warning system: case study in Kaijuri union, Bangladesh. Int J Disast Risk Reduct.
is not the ordinary weather conditions, but the intense or rare weather 2015;14(Part 4):323–31 [ISSN 2212-4209].
events, that produce the greatest societal impacts. [26] GAR. Cost and benefits of early warning systems by David Roger and Vladimir
Tsirkunov; 2011 [Global Assessment Report of UNDRR].
Weather forecast services are traditionally based on conventional syn-
[27] GCCE. Land and living systems. Retrieved from Global Climate Change Explorer https://
optic methods, enhanced by the use of numerical weather prediction. How- www.exploratorium.edu/climate/land-and-living-systems; 2020.
ever, there is a growing demand for predictions of weather and climate [28] GCF. Strengthened weather and climate services for resilient development for Pacific
Islands. Green Climate Fund; 2018.
conditions that go beyond the usual weather forecast window (days 1–7),
[29] Gibson PB, Waliser DE, Goodman A, DeFlorio MJ, Delle Monache L, Molod A.
with interest in medium-range outlooks extending from weeks to months. Subseasonal-to-seasonal hindcast skill assessment of ridging events related to drought
This drives an emerging shift from single deterministic model weather fore- over the western United States. J Geophys Res Atmos. 2020;125:e2020JD033655.
cast outputs, to probabilistic forecasts generated from ensemble prediction https://doi.org/10.1029/2020JD033655.
[30] Giuseppe FD, Vitolo C, Krzeminski B, San-Miguel J. Fire weather index: The skill pro-
systems, covering the medium-range window (7–30 days). vided by ECMWF ensemble prediction system. Natural Hazards Earth Syst Sci. 2020;
20:2365–78.
[32] GoUK. National risk register. https://www.gov.uk/government/publications/national-risk-
Disclaimer register-2020; 2020.
[31] Gluckman P, Bardsley A. Uncertain but inevitable: The expert policy political nexus and
high impact risk. Koi Tu: The Centre for Informed Futures. University of Auckland. New
The article does not reflect the official views of the UN. Zealand; 2021.
[33] Griffiths G. Ensemble forecasting. Retrieved from MetServices Blog https://blog.
metservice.com/; 2015.
[34] Helen Fischer, et al. Environ Res Lett. 2020;15:114041. https://doi.org/10.1088/1748-
Declaration of Competing Interest 9326/abbc3c.
[35] Hopson T, Webster P. A 1–10-day ensemble forecasting scheme for the Major River
basins of Bangladesh: forecasting severe floods of 2003–07. J Hydrometeorol. 2010;
None. 11(3):618–41 Retrieved August 17, 2021, from http://www.jstor.org/stable/24912
857.
[37] IFRC. Forecast-based financing. Retrieved from https://www.forecast-based-financing.
Acknowledgement org/; 2021. International Decade for Natural Disaster Reduction.
[38] IPCC. Climate change 2014: Synthesis report. Contribution of working groups I, II and
The authors would like to thank Tonkin + Taylor in partnership with III to the fifth assessment report of the intergovernmental panel on climate change
[Core writing team, R.K. Pachauri and L.A. Meyer (eds.)]. Geneva, Switzerland: IPCC;
Integrated Research on Disaster Risk (IRDR), Committee on Data 2014. p. 151.
(CODATA) of the International Science Council (ISC) for organising the [40] Klinke A, Renn O. The coming of age of risk governance. Risk Anal. 2021;41:544–57.
webinar on “Create resilient communities with medium-range hazard https://doi.org/10.1111/risa.13383.
[42] Lena Weingärtner, Jaime Catalina, Todd Martin, Levine Simon, McDowell Stephen,
warning systems”. Authors also thank Neeraj Shankar of Tonkin + Taylor MacLeod Dave. Reducing flood impacts through forecast-based action. Working Paper
for providing editorial support. 553; 2019.
[45] Met Office. Causes of extreme fire weather in Australia. Retrieved from. Retrieved from
Met Office https://www.metoffice.gov.uk/research/news/2020/causes-of-extreme-fire-
References
weather-in-australia; 2020.
[47] NEMA. National Disaster Resilience Strategy. Wellington: Ministry of Civil Defence &
[5] Anticipation Hub. Anticipation Hub. Retrieved from https://www.anticipation-hub.org/ Emergency Management; 2019.
about; 2021. [48] Nicholas Simpson, Mach Katharine J, Constable Andrew, Hess Jeremy, Hogarth Ryan,
[6] Bapon Fakhruddin, Clark Helen, Robinson Lisa, Hieber-Girardet Loretta. Should I stay or Howden Mark, et al. A framework for complex climate change risk assessment. One
should I go now? Why risk communication is the critical component in disaster risk re- Earth. 2021;4(4):489–501 [ISSN 2590-3322].
duction; 2020 [Progress in Disaster Science]. [49] NIWA. Forecasting weather systems. Retrieved from National Institute of Water and At-
[7] Battiston S, Caldarelli G, May RM, Roukny T, Stiglitz JE. The price of complexity in fi- mospheric Research-Taihoro Nukurangi https://niwa.co.nz/climate/research-projects/
nancial networks. Proc National Academy of Sciences USA, 113. ; 2016. p. 10031–6. forecasting-weather-systems; 2020.
https://doi.org/10.1073/pnas.1521573113. [50] NZIER. The value of MetService’s public weather forecasts and weather warnings. New
[8] Bjerknes V, Hesselberg TH, Devik O. Dynamic meteorology and hydrography, pt.H; Ki- Zealand: New Zealand Institute of Economic Research; 2018.
nematics, Carnegi e Inst. Washington Pub. no. 88, pt. II; 1911; 175. [52] Panteli Mathaios, Mancarella Pierluigi. Influence of extreme weather and climate
[9] Bosetti V, Cooke R, Dutt V, Ha-Duong M, Held H, Llanes-Regueiro J, et al. Integrated risk change on the resilience of power systems: impacts and possible mitigation strategies.
and uncertainty assessment of climate change response policies. Climate change 2014: Electric Power Syst Res. 2015;127(2015):259–70.
mitigation of climate change: working group III contribution to the fifth assessment re- [53] Privé Nikki C, Errico Ronald M. The role of model and initial condition error in numer-
port of the intergovernmental panel on climate change. Cambridge University Press; ical weather forecasting investigated with an observing system simulation experiment.
2014. p. 151–2. Tellus A: Dynam Meteorol Oceanogr. 2013;65:1. https://doi.org/10.3402/tellusa.
[10] Buizza R. Introduction to the special issue on 25 years of ensemble forecasting. Q J Roy v65i0.21740.
Meteorol Soc. 2019;145(Suppl. 1):1–11. https://doi.org/10.1002/qj.3370. [54] Renn O, Lucas K, Haas A, Jaeger C. Things are different today: the challenge of global
[11] Burkholz R, Garas A, Schweitzer F. How damage diversification can reduce systemic systemic risks. J Risk Research. 2019;22:401–15. https://doi.org/10.1080/13669877.
risk. Phys Rev E. 2016;93:042313. https://doi.org/10.1103/PhysRevE.93.042313. 2017.1409252.
[12] CARICOM. National strategic plan and framework for weather. Water and climate ser- [56] Schroeter S, Richter H, Arthur C, Wilke D, Dunford M, Wehner M, et al. Forecasting the
vice for 8 CARICOM countries. Caribbean: Meteorological Organization (CMO); 2021. impacts of severe weather. Australian Institute for Disaster Resilience; 2021.
[13] Carnaghan Carla. Business process modeling approaches in the context of process level [58] Seng DC. Improving the governance context and framework conditions of natural haz-
audit risk assessment: An analysis and comparison. Int J Account Inform Syst. 2006;7 ard early warning systems. J Integr Disast Risk Manag. 2012;2(1). https://doi.org/10.
(2):170–204 [ISSN 1467-0895]. 5595/idrim.2012.0020.
[14] CIFDP. Coastal inundation forecasting demonstration project. World Meteorological Or- [59] Sofyan Sufri FD. A systematic review of community engagement (CE) in disaster early
ganization (WMO); 2019. warning systems (EWSs); 2020. https://doi.org/10.1016/j.pdisas.2019.100058 Prog-
[15] Cloke HL, Pappenberger F. Ensemble flood forecasting: a review. J Hydrol. 2009: ress in Disaster Science.
613–26. [60] SPREP. Pacific Islands meteorological strategy 2012–2021 - sustaining weather and cli-
[16] Convertino M, Annis A, Nardi F. Information-theoretic portfolio decision model for op- mate Services in Pacific Island Countries and Territories. Samoa: Secretariat of the Pa-
timal flood management. Environ Model Software. 2019;119:258–74. https://doi.org/ cific Regional Environment Programme; 2012.
10.1016/j.envsoft.2019.06.013. [66] White Christopher J, et al. Potential applications of subseasonal-to-seasonal (S2S) pre-
[18] CRED. Human cost of disasters. Centre for Research on the epidemiology of disasters. dictions. Meteorol Appl. 2017;24(3):315–25.
United Nations: Office for Disaster Risk Reduction; 2020. [67] Willams TM. Using a risk register to integrate risk management in project definition. Int
[19] Demeritt D. Challenges in communicating and using ensembles in operational flood J Project Manag. 1994:17–22.
forecasting. Meteorol Appl. 2010:209–22. [68] Witze A. Improved three-week weather forecasts could save lives from disaster. Re-
[21] Eurpoean Commission. European civil protection and humanitarian aid operations - Pa- trieved from Science News https://www.sciencenews.org/article/climate-weather-
cific region; 2020 [Eurpoean Commissio]. forecast-three-week-disaster-storms; 2020.

7
B.(S.H.M.) Fakhruddin et al. Progress in Disaster Science 12 (2021) 100203

[69] WMO. Subseasonal to seasonal prediction project: Bridging the gap between weather meteorological service and the global Facility for Disaster Reduction and Recovery;
and climate. WMO Bull. 2012;61(2) World Meterological Organisation. 2016 [Retrieved from htt].
[70] WMO. Impact-based forecasting and warning: weather ready nations. Retrieved from [76] Yudkowsky E. Cognitive biases potentially affecting judgment of global risks. In:
https://public.wmo.int/en/resources/bulletin/impact-based-forecasting-and-warning- Bostrom Nick, Ćirković Milan M, editors. Global catastrophic risks. Oxford University
weather-ready-nationsWorld Meteorological Organisation; 2018. Press; 2008. p. 91–119.
[71] WMO. Multi-hazard early warning systems: a checklist. Retrieved from https://library. [77] Zsoter E, Prudhomme C, Stephens E, Pappenberger F, Cloke H. Using ensemble
wmo.int/doc_num.php?explnum_id=4463World Meteorological Organisation; 2018. reforecasts to generate flood thresholds for improved global flood forecasting. J Flood
[72] WMO. Natural hazards and disaster risk reduction. Accessed 25th November 202 Risk Manag. 2020;13(4).
https://public.wmo.int/en/our-mandate/focus-areas/natural-hazards-and-disaster-risk- [78] Fakhruddin Bapon, Mudasser Khan. Medium Range Ensemble Flood Forecasting in WAi-
reductionWorld Meteorological Organisation; 2020. kato River Basin, New Zealand. 2018 HEPEX Workshop, Breaking the barrier, 6-8 FEb-
[73] WMO. The future of forecasts: impact-based forecasting for early action. Retreived from ruary 2018; 2018 Melbourne, Australia https://hepex.inrae.fr/wp-content/uploads/201
https://www.forecast-based-financing.org/wp-content/uploads/2020/09/Standalone- 8/02/HEPEX_workshop_proceedings.pdf.
Impact-based-forecasting-guide-2020.pdfWorld Meterological Organisation; 2020. [79] Global Assessment Report on Disaster Risk Reduction. United Nations; 2015https://
[74] World Bank. Implementing multi-hazard impact-based forecast and warning services: a www.preventionweb.net/english/hyogo/gar/2015/en/home/index.html.
report on a workshop organized by China Meteorological Administration – Shanghai [80] Rahman Alfi, Munadi Khairul. IOP Conf. Ser.: Earth Environ. Sci., 273; 2019 012040.