AI for Natural Disaster Management
Summary
TLDRDr. Monique Kudlish, Innovation Manager at the Fraunhofer Institute of Telecommunication and chair of the ITU-WMO Focus Group on AI for natural disaster management, discusses the potential of AI in mitigating natural disaster risks in Africa. She highlights the importance of leveraging satellite and ground data for early warnings and emphasizes the need for accessible information to prevent acute food insecurity from becoming chronic. The conversation also touches on the challenges of data accessibility and the role of AI in various stages of disaster management, including preparedness and response.
Takeaways
- 🌍 Natural disasters, such as floods, droughts, and storms, are particularly impactful in Africa due to their wide-ranging effects on health, property, and the environment.
- 📈 The loss of agricultural production due to natural disasters is estimated to affect 2-4% of Africa's GDP, highlighting the importance of effective disaster management.
- 🤖 AI can enhance the prediction and mitigation of natural disasters in Africa by leveraging satellite data and machine learning to provide early warnings and improve decision-making.
- 🌡️ Projects like NASA's Harvest program combine earth observations and weather forecasts with AI to offer early warnings, especially beneficial for smallholder farmers.
- 📊 The accessibility of data is crucial, but it often requires processing and understanding metadata to be effectively used for forecasting and disaster management.
- 🏞️ In Africa, the impact of natural disasters on health can be both direct, such as heat stroke or asphyxiation, and indirect, like waterborne illnesses or food shortages.
- 🔄 AI's role in disaster management includes all stages of the cycle, from preparedness and response to recovery, with a focus on data collection, modeling, and effective communication.
- 🌐 The ITU-WMO Focus Group on AI for Natural Disaster Management aims to create a roadmap, conduct workshops, and produce technical reports to enhance the use of AI in this field.
- 📱 The potential of smartphones in AI applications should not be underestimated, as they can be used for various tasks, including identifying pests or health issues in resource-limited settings.
- 🏛️ Governments and NGOs in Africa are encouraged to engage with the focus group to improve natural disaster preparedness and response, utilizing AI and data analysis.
Q & A
What is the role of Dr. Monique Kudlish in the field of AI for natural disaster management?
-Dr. Monique Kudlish is the Innovation Manager at the Fraunhofer Institute of Telecommunication in Berlin, Germany, and she is also the chair of the newly established ITU-WMO Focus Group on AI for Natural Disaster Management.
How does Dr. Kudlish define natural disasters?
-Natural disasters are defined as damaging physical events of a predominantly natural origin, which can include atmospheric, hydrologic, geophysical, oceanographic, and biologic events. They can lead to injury, mortality, displacement, property damage, and disturbance to natural resources.
Which natural disasters are most impactful in Africa according to the podcast?
-On a continental scale in Africa, epidemics, floods, droughts, and storms are considered the most impactful natural disasters in terms of impacts and mortality.
How does climate change affect food and nutrition security in Africa?
-Climate change affects food and nutrition security in Africa by impacting agricultural production, which is estimated to cause a loss between two and four percent of GDP.
How can AI help African farmers predict natural disasters and forecast agricultural production?
-AI can leverage ground and remotely sensed data, such as satellite imagery, to provide insights into the mechanisms of natural disasters and improve forecasting and warning systems for such events, helping farmers make informed decisions and adaptation plans.
What is the NASA Harvest program and how does it assist African farmers?
-NASA's Harvest program works with earth observations and weather forecasts combined with machine learning methods to provide early warnings, particularly for smallholder farmers in Africa, to help them prepare for natural disasters.
What challenges does Africa face in terms of accessing and utilizing AI for natural disaster management?
-Africa faces challenges such as ensuring that information about natural disaster threats is accessible to all involved in the food system, including decision-makers, farmers, and traders. Additionally, the continent's diverse and small-scale farms require customized adaptation strategies.
How does the availability and accessibility of data from sources like NASA impact AI's role in natural disaster management?
-Data from sources like NASA is publicly available but may require processing and understanding of metadata and data limitations before it can be effectively used in AI models for natural disaster management.
What types of data are useful for AI models in predicting and managing natural disasters?
-Useful data for AI models includes satellite data, drone imagery, instrumental data from weather stations and river gauges, and crowdsourced data. These can be integrated into AI models to enhance prediction and management of natural disasters.
How does Dr. Kudlish see AI mitigating the risk associated with the effects of climate change in Africa?
-Dr. Kudlish believes AI can provide insights into the mechanisms behind hydrometeorological hazards, help detect disasters, and offer more accurate forecasts, which can lead to more informed decisions and potentially reduce the costs associated with such events.
What is the ITU-WMO Focus Group on AI for Natural Disaster Management working on, and how can stakeholders get involved?
-The ITU-WMO Focus Group is working on building a roadmap of AI activities, organizing workshops, creating technical reports, and developing educational materials. Stakeholders can get involved by participating in these activities and engaging with the focus group to improve natural disaster preparedness.
Outlines
🌐 Introduction to AI in Natural Disaster Management
Dr. Monique Goodluck, Innovation Manager at the Fraunhofer Institute of Telecommunication and Chair of the ITU-WMO Focus Group on AI for natural disaster management, discusses the application of AI in mitigating natural disaster risks. She explains that natural disasters, such as floods, droughts, and storms, have significant impacts on Africa, affecting food and nutrition security and causing economic losses. AI can help predict these disasters and forecast agricultural production by leveraging satellite data and machine learning, aiding in decision-making and adaptation planning for farmers and governments.
📊 Accessibility and Application of AI in African Agriculture
The conversation delves into the accessibility of data for AI applications, particularly in African agriculture. While much data is publicly available, it often requires processing and understanding to be useful. Dr. Goodluck emphasizes the need for accessible information on threats to all stakeholders in the food system. She acknowledges the diversity in African farms and the importance of tailored adaptation strategies. The focus group's work aims to explore how AI can provide hazard information and improve forecasting and communication of events to enable informed decision-making.
🌡️ AI's Role in Disaster Management Cycle
Dr. Goodluck outlines AI's potential at various stages of the disaster management cycle, focusing on preparedness and response. The discussion covers data collection and monitoring, modeling for forecasting events, and effective communication through early warning systems. High-quality data is essential for AI models, and the focus group is investigating how AI can enhance data quality, support real-time detection, and identify complex patterns. The goal is to improve natural disaster preparedness and response using AI and data-driven insights.
🌍 AI's Potential in Climate Change Adaptation in Africa
The podcast addresses how AI can help mitigate the risks associated with climate change in Africa, where hydrometeorological hazards are intensifying. AI can provide insights into hazard mechanisms, improve detection and forecasting, leading to more informed decisions on land development, crop selection, and evacuation strategies. Dr. Goodluck expresses hope that AI can reduce the costs of such events and benefit communities, especially in low-resource contexts.
🤝 Engaging Stakeholders in AI for Disaster Management
Dr. Goodluck invites engagement from various stakeholders, including governments, NGOs, and communities, in the focus group's efforts to utilize AI for disaster management. The group aims to create a roadmap, conduct workshops, and produce technical reports and educational materials. She stresses the importance of African expertise and data in building relevant AI models and encourages participation from the African community to ensure the focus group's work is impactful and context-specific.
Mindmap
Keywords
💡Artificial Intelligence (AI)
💡Natural Disasters
💡Climate Change
💡Remote Sensing
💡Data Privacy
💡Machine Learning
💡Disaster Management Cycle
💡Early Warning Systems
💡Agricultural Production
💡Focus Group
Highlights
Dr. Monique Goodluck discusses the potential of AI in mitigating natural disaster risks.
Natural disasters are defined as damaging physical events of predominantly natural origin.
In Africa, epidemics, floods, droughts, and storms are the most impactful natural disasters.
AI can leverage satellite data to provide insights into natural disaster mechanisms and forecasting.
NASA's Harvest program combines earth observations and machine learning to provide early warnings for farmers.
AI can help in making informed decisions and adaptation plans to reduce the impact of natural disasters on agriculture.
Data from satellites and other sources are publicly available but may require processing for practical use.
Dr. Goodluck emphasizes the need for accessible and understandable data for all stakeholders in the food system.
AI can provide information about natural disaster mechanisms and improve forecasting and communication.
Natural disasters can have both direct and indirect impacts on health, including exposure to elements and infrastructure collapse.
AI and big data can intervene at each stage of the disaster management cycle, from preparedness to response.
The focus group on AI for natural disaster management aims to build a roadmap and engage researchers and stakeholders.
Dr. Goodluck highlights the importance of engaging African communities and stakeholders in AI research for natural disaster management.
AI has the potential to reduce the costs associated with natural disasters by providing better predictions and insights.
The focus group is an open community for stakeholders and experts to work together on AI in natural disaster management.
Dr. Goodluck invites interested parties to join the focus group and contribute to the discussion on AI and natural disasters.
Data privacy and confidentiality are important considerations when collecting data from various sources, including social media.
Transcripts
hello everyone welcome to this podcast
brought to you by efi the african forum
of artificial
artificial intelligence for help my
guest today is dr monique
goodluck she's the innovation manager
at friend offer institute of
telecommunication berlin
germany and chair of the newly
established
itu wmo
focus group on ai for natural disaster
management
he was the lead technical editor for
seven journals at the american
meteorological society
and has conducted research on
past climate change extreme weather
events and regional climate model
project
projections here today to talk about
how ai can be applied to mitigate the
risk of natural disaster
dr kudlish good afternoon hello good
afternoon thank you for the introduction
can you tell us what other natural
disasters how can we define it
certainly so natural disasters are
damaging physical events of a
predominantly natural origin
so they can be atmospheric hydrologic
geophysical
oceanographic biologic of these types of
origins
impacts can include everything from
injury to mortality
displacements damage to property or
cultural heritage
infrastructure and also disturbance to
nature natural resources
okay very good now all those natural
disasters which one
are more relevant to africa
yeah so i mean obviously africa is a big
continent so depending on where you are
there are going to be different types of
of hazards that are are more relevant
but um i think on a continental scale
in terms of impacts and mortality
epidemics
floods droughts and storms are are the
most
impactful okay
now it's worth mentioning that in africa
food and nutrition security are affected
by climate
and environmental assets the loss of
agricultural production is estimated
between
two and four percent gdp uh how do you
think ai
can help african farmers uh better
predict
uh natural disasters and and better
forecast
uh agricultural production yeah
um i mean indeed natural disasters
whether
droughts floods or the locust plagues
that we saw last summer in eastern
africa
can absolutely threaten crop
productivity and
disrupt supply chains which can cause
acute food insecurity
through ai we can leverage both ground
and remotely sensed
so satellite data to get insight into
the mechanisms of these hazards
as well as our ability to forecast and
warn for such disasters
so there are projects like nasa's
harvest program which works with
within this space combining earth
observations
weather forecasts with machine learning
methods
and providing early warnings in
particular for smallholder farmers in
africa
and this type of information can help us
make more informed decisions and
adaptation plans
so things like uh you know what are the
best farming practices that i can do
to reduce the impacts of such hazards on
the productivity in
in my farm or how can governments
mitigate the impacts of such hazards
using targeted relief
so this is something that uganda's
disaster risk financing fund has done
using early warning systems so that they
can really focus their relief
in in a targeted way there are some
challenges though as you can imagine um
we need to make sure that this
information about these threats
is accessible to those who are involved
in the food system
so decision makers farmers traders
everyone within this chain needs to have
access to this information
so that this acute food insecurity
doesn't
become chronic we also need to consider
that
particularly in sub-saharan africa there
are many small and diverse farms
each one has their own array of crops
and own
agricultural practices so adapting
to this type of event is going to vary
from farm to farm
so within the context of the focus group
we're exploring how ai can provide
information
about the mechanisms of these hazards
and how we can improve our ability to
forecast and communicate this type of
event
so that individuals in this situation
are able to use that information to make
decisions
interesting now since you're speaking
about uh data
i mean is it accessible uh is it easy to
get uh data let's say from nasa
or do you or do farmers need to go
through
some kind of bureaucracy
just to get the data they need to uh
you know that they're going to need to
forecast their production
yeah no it's a great question so a lot
of these data are publicly available
um they're maybe not so user-friendly
because they're gonna have imagery
you need to calibrate them you need to
know what you're looking for
um so i don't know um
how much someone needs to be aware of
these aspects of the data in order to
leverage them for this um
but you know obviously researchers who
are in this space are familiar with
these practices
so um they are of course able to access
the data and use them
okay so basically you need to kind of
process the data before you can actually
use it
yeah exactly you need to understand the
metadata you need to understand you know
the best way to use this data you need
to understand the sort of limitations of
the data
what to look for when it comes to data
quality so that it can be best used
okay i see um
now since you've been doing research
with some of the african researchers um
did you notice any um
difficulties uh when it came to africa
as far as using the data uh does africa
isn't kind of platform
or any kind of infrastructure i would
say
or ecosystem where we can actually use
the data
to apply to agriculture
or any other sectors that might need
data when it comes to ai
or for natural disasters so to be honest
i'm not
that far in the focus group to answer
that question it's a great question and
i look forward to exploring that
um so we've just kicked off the focus
group we're just trying to
engage researchers and stakeholders from
africa and our activities
um so i can't yet comment on that but
it's a good question i'm of course
curious to know if there are
you know platforms with benchmarking
data in africa and and you know how we
can access those so that's definitely
something that i'll be looking into
okay and as far as data collection on
the field in africa
um i know previously you mentioned you
know we need
weather uh stations uh different places
uh obviously that's the important data
that we need uh
what does my need might be needed on the
ground to make
ai work for natural disasters
yeah so i mean there's so many different
types of data that could be
integrated into an ai model um
and often there are times types of data
that you wouldn't even know would have
value but you
test it out and you find out that there
is some sort of a synergy or some sort
of relationship
so um i mean satellite data
are definitely a very important source
of information especially in this type
of
activity because you can access data
that
logistically would be difficult to get
otherwise also
data from drones so aerial imagery
from the ground you know instrumental
data whether
you know gauges and rivers or weather
stations
um are of course interesting and then of
course crowdsourced data can can be used
okay as far as the impact on health
how does natural disaster impact health
yeah that's a great question um it
definitely depends on the type of event
and vulnerability um but you can have
both
direct impacts and indirect impacts so
direct impacts um would be things
like through exposure to the elements a
heat stroke during a heat wave or
you know asphyxiation from forest fires
you know this would be
something that directly causes injury or
death um
also collapse of infrastructure you know
we saw with the
tsunami um and fukushima you can have
secondary disasters as well
um you can be struck by debris or
involved in traffic accidents these are
also sort of the what i would say are
direct impacts and then you also have
these
indirect impacts um contamination to
water supplies can spread
you know waterborne illness um pathogens
can be transported we've seen that
um with dust storms carrying fungi that
cause valley fever
um power outages we just had this in the
u.s
with the cold snap in you know texas
that you know people
you know suffered because they weren't
able to get heating um
of course you know food shortages other
disruptions to the supply chain
can can cause um consequences
or damage to hospitals
communications disruptions these can all
have indirect impacts on health
i see yeah there's a lot to consider
there i know
um now as as we know that disaster
management
management involves uh several stages uh
can you explain how
ai and big data intervene at each stage
of disaster management
so yeah so i mean as you say you know
there's the disaster management cycle
and of course
at all phases of the cycle there are
ways that ai can benefit
um i'm gonna sort of bring it back to
the focus group um
within the focus group we're looking at
a slice of the disaster management cycle
which is in the area of preparedness and
response so
we're looking at uh the potential and
also pitfalls
of using ai when it comes to data
collection
monitoring data handling so that's like
the first pillar
the second pillar is modeling so
forecasting events or also
reconstructing events
and then effective communication so
early warning systems this type of
communication um so i mean as
you know you and i know high quality
data are really the foundation for
ai models they're also the foundation
for understanding hazards and
the underlying mechanisms um they're
needed when it comes to providing ground
truth so understanding what you're
seeing in satellite image
calibrating models and and building
these
reliable ai algorithms so things in the
focus group that we're going to ask
in this space are how can ai be used to
enhance
data quality and data quantity how can
ai support
the detection of features in real time
so if you have a seismic time series how
can we
detect you know in real time that an
event is happening
and how can we use it to identify
complex relationships
and patterns within data
when it comes to modeling we're looking
at um how
these algorithms can enhance and improve
traditional models
so physics based or numerical based
models
and some questions that we could look at
are what's the current gold standard
method when it comes
to making a forecast or a reconstruction
and how can these algorithms
bring that to the next level what
requirements should data meet when we're
training and testing
this type of algorithm and what should
we consider when we evaluate an
algorithm
what metrics should we use what
expectations should we have when it
comes to explainability or transparency
and then within effective communication
so this would be this sort of response
part of the um cycle
we might consider once a disaster has
been forecast or triggered how can ai
assist with
creating an early warning with sending a
push message to a phone or
using automated translation services how
do we ensure that communications methods
are reliable and trusted by the
population
are they accompanied by protocols to
ensure that individuals know how to
respond to these messages
so again this is just a slice of the
disaster management cycle of course
ai could be used in other areas but this
is where we're focusing for the focus
group
okay
all right um
in africa the effect of climate change
as uh
that's where exacerbated the frequency
and intensity of floods
in some areas uh like mozambique a
couple years ago they had a huge flood
there
uh water everywhere a lot of people got
displaced
um and then the northern part of africa
is
impacted by droughts you know countries
like um china and
ethiopia um and drc and kenya had issues
with
heavy rain and and and flood lands
slides as well
with all those natural disasters
happening due to
climate change and global warming uh
how do you see ai uh mitigating
the risk of associating the freight of
climate change
uh in this part of the world yeah
so i mean unfortunately such
hydrometeorological hazards
have been intensifying and will probably
continue to intensify
um in many regions of the world as a
result of climate change
um it's also related to you know where
we're settling where
our urbanization is happening um and and
the way these factors interact
um i mean through giving insight into
the mechanisms behind these hazards
helping us detect
disasters and giving more accurate
forecasts
ai is going to let us make more informed
decisions
so where should i consider to develop
land what crops should i
choose to use because they're more
resistant to this type of event
um when an event is happening you know
how
and where and when should we evacuate
and what's the best way to do so
so it's my hope that through having this
extra insight we can
reduce the costs of such events
there's definitely a cost benefit
associated with ai
now for the low context of africa low
resource context
obviously africa need technology
that's pretty much known how do you
think we can maybe
vulgarize ai to where it's going to be
accessible
to smaller communities like villages
especially when it comes to disaster
management
what kind of model are you saying
that we can we can use in africa to
disseminate ai
to where even you know the smaller
communities that already have high tech
uh can benefit from ai and and prevent
natural disasters in their
smaller communities yeah what can be
done
in that area you think yeah it's a good
question so i mean i think on the one
hand there's the question
i mean when it comes to you know
building the models
so obviously you know there are such
great research happening in africa
of course to build models you need to
have a certain amount of computing
capability and capacity so that's
obviously
you know something that needs to be
considered uh you know these types of
resources
so that the research can be done within
africa using african data and
african expertise um when it comes to
actually using the outcome of these
this research so the model um i mean
we've seen from for example from the
focus group on ai for health
you know with a smartphone you can do
incredible things
you know detecting
you know skin lesions
using your phone uh identifying snakes
species um in the field so i think that
um
we shouldn't underestimate what a
smartphone can do because those are
uh quite available to
i would assume also people who are you
know
using farms um so if they you know have
a pest or or something else that they
want to identify i would imagine that
this would be something that they could
use
very good um what about the government
and ngos obviously they have more
resources than smaller communities
what can they do to get more involved in
ai
and and and better prepare for natural
disasters uh in the african context
you think yeah i mean that's a great
question so
um our research in the focus group
we're gonna be working on four kinds of
deliverables
um we're going to be building a road map
of ai activities
within this area of natural disaster
management
we're going to be having workshops that
are going to be bringing together
experts and stakeholders
and highlighting activities within this
framework
and i should mention that our kickoff
workshop is going to be this monday
from 10 a.m to 2 p.m west africa time
and you can register on our website um
and then we're also gonna have meetings
the following two days tuesday and
wednesday
um we're also going to be making uh
technical reports to summarize the key
findings of our analyses based on
selected use cases and educational
materials to support capacity building
now for these deliverables to have value
for african stakeholders it's really
important that we have engagement from
the region i want to know
what use cases are most relevant for the
african community i want to know
what research is being done in the
african community i want that to be in
our
analysis and in our report um i know
that there's really exciting work being
done
on the continent um in the past few
weeks i've had the pleasure of speaking
with
representatives from the disaster risk
reduction group at the african union um
with the vice chair from the african
science and technology advisory group
with volcanologists from the goma
volcano laboratory and drc
and researchers from strathmore
university in kenya and of course i also
know through
the other focus group on ai for health
many researchers in africa that are
doing amazing research in this space
so this has all shown me that
you know africa is a very fertile ground
when it comes to using ai
so you know there's no reason that we
shouldn't have really active engagement
within this focus group
we have abundance of data we have top
tier machine learning experts
and there's an interest in the
application so i think that that's going
to be really key when it comes to
bringing in
the communities governments the ngos
into the focus group so that they can
improve
natural disaster preparedness okay
very good dr kuvlich
thank you pretty much answer all our
questions i don't know if you want to
add
anything uh that you think might be
relevant for this podcast
um as far as ai and how it can you know
help prevent natural disasters maybe
that we need to cover
um well i i would just love to sort of
loop back to the focus group
and and just you know encourage anyone
who's interested in ai
who's interested in in data who's
interested in
how this can be used to support natural
disaster management anyone from the
natural disaster or drr
space the focus group is really an open
community
it's a platform for us all to work
together on this topic
whether your expertise is on the natural
disaster spectrum
disastrous reduction spectrum or the ai
spectrum
this is really a space for stakeholders
experts
policy makers researchers everybody to
get together
and to to tackle this head on so i
really
hope that anyone with an interest in the
topic will
visit the focus group website i don't
know if i can
put it in the chat maybe that's the
easiest
i actually have a few questions from the
the chat so that
i'm gonna ask just came in stick it in
the chat so that anyone can see it and
yes
please there we go uh
i have a question about the collection
of data um
what kind of issues are you seeing
with the collection of data when it
comes to trying to get the data from
social media do you see any kind of
confidentiality
issues uh with that or is it pretty much
um
you know open for everybody to get get
whatever data they want to get
from platform like um
facebook or something else uh
is it pretty easy to get data from those
uh platforms or not
that's a really great question so like i
said we're sort of at this
nascent stage of the focus group so we
haven't even gotten to that level yet
of collecting data we're just building
up a community
and this is definitely something that
we're going to look into i mean also
when it comes to
you know imagery from from drones or or
imagery from satellites i mean there are
also
you know data privacy issues that we
need to keep into consideration
um so this is definitely something
that's on my mind and it's definitely a
great question but
within the the focus group we're not at
that level yet where we were addressing
that but it will be something that we
look into
okay all right sounds good i don't have
any other questions for you
um thanks a lot for your time uh this
was
very good a great podcast
i wish you all the luck with uh your fox
group thank you very much thank you and
uh for the rest of the audience uh this
podcast is gonna be archived
on uh the african form of artificial
artificial intelligence
website the address is www.afaih.com
thank you very much take care bye-bye
Ver Más Videos Relacionados
Geografi Kelas XI (28) Siklus Penanggulangan Bencana | Pra Bencana, Tanggap Darurat, Pemulihan
¿Qué hay detrás de los APAGONES DE INTERNET? La Ciencia de las Redes que no se Caen ante Desastres
Interview with Kagawad Nelson Brila
MGA HAKBANG SA PAGBUO NG COMMUNITY BASED DISASTER RISK REDUCTION AND MANAGEMENT PLAN Video Lesson
FEU Public Intellectual Lecture Series | Dr. Mahar Lagmay | Part 2
AWANI Review: AI-Driven Future: Insights from Amazon’s CTO
5.0 / 5 (0 votes)