Tip:
Highlight text to annotate it
X
Great East Japan Earthquake Big Data Workshop Project 311 A Visualization of the Evacuation Process During the Tsunami Team Masters & Forever 22
This is the presentation for Team Masters & Forever 22.
I am Lena Takayasu, Tokyo University.
This is my first time here.
What we examined in our project is the question
of why the evacuation during the tsunami was slow and late.
Many approaches look at the confusion in the cities of the disaster region.
Because we thought it might save lives during a future disaster,
we instead looked at the stream of people fleeing the tsunami on the coast.
Looking at changes in population density in the 250m squares via the ZENRIN data,
we decided to visualize them with Google Maps, in conjunction with shelter and flooding data.
I would like to present the results we deemed most important in the event of a future disaster.
We picked the town of Otsuchi in Iwate Prefecture as our model case.
The reason is detailed in writing here.
This is a video of the ZENRIN population density data from the day of the quake
visualized through Google Maps. The red line is the arriving tsunami,
we can see municipal buildings and a hospital alongside the river,
and that people congregated there during the day.
The earthquake happened at 2:46 p.m. The Meteorological Agency issued
an urgent report that the tsunami is estimated at 3m in Iwate.
The Otsuchi levees are 6.4m tall, so many people figured that they wouldn't have to move,
and there are even people who took cameras to the top of the levees to look at the sea.
This next picture is from around 4 p.m. Most people have already moved away from the coast.
About 30 minutes after the quake the Meteorological Agency issued its first correction of the initial estimate.
At this point the first wave reached the town.
We can see that in Otsuchi, the mere 30 or so minutes it took
for the waves to surmount the levees and enter the town would decide a person's fate.
It'd seem that, with Otsuchi being a small town, everyone should be able
to reach an area of sufficient height to be safe within about 10 minutes.
However, this is the route an actual tsunami victim took to evacuate.
We can see that the route is pretty roundabout.
It would be good if we could follow the movements of people
in these decisive 30 minutes in greater detail.
For example, if we could see patterns of people who didn't make it, we could use
that knowledge to prevent something similar during a future disaster, I think.
After the tsunami's arrival we don't see a large fluctuation
of people at once, but rather a slow decrease.
As was said earlier, this data is based on the GPS information from cell phones.
So it is likely that because of the power outage
people eventually ran out of battery and lost their GPS connection.
As a result we can't trace their movements anymore.
Let's look at the last data we have in detail.
On the left we see population distribution from 2 to 3 a.m. on the day after the quake,
and on the right, from 3 to 4 a.m. This is the last data.
Looking at this image, there are a lot of areas showing up as 190,
although it's kind of hard to see here.
There are several other numbers in between,
but most people have disappeared in the graph on the right.
Basically what this means is that these numbers,
such as the ones in the orange-outlined area, are derived via a process like so.
We have the number 92, for example, signifying people who lost reception after 29 minutes.
This is about the number we would expect by multiplying 29/60 by 190.
So the numbers represent the proportion of people with reception over time.
I think that this kind of process is likely the result of data anonymization.
Because of this process we unfortunately can't trace
the changes in the original population anymore.
It is rather difficult to reconstruct the evacuation from this data.
But we do learn two important things from looking at it.
For 12 hours after the tsunami we have multiple
cell phone signals even from flooded areas.
And also we know how many hours and minutes individual cell phones lasted.
We can discover several useful ways of using cell phones
in the event of a disaster from this data.
The first is to put phones into energy-saving mode,
which can make the battery last up to three days.
This will stop unnecessary functions and applications,
and slow GPS transmissions down to about 1 in 5 minutes.
Cell phones lose battery power much quicker when reception is bad.
But even when there is no reception at all, like during the disaster,
we can still trace the owner's whereabouts for three days with the GPS.
If we could relay personal information like that to firefighters
or the Self-Defense Force or relatives, it would help the search effort.
Also, if we had cell phone relay stations that could be transported by helicopter,
then that would enable us to send evacuation
and rescue information via e-mail to people in need.
But of course in order to do something like that privacy laws would have to…
[gong; unintelligible]
...important to use cell phone data to rescue disaster victims.
This is all on our posters, so if you are interested please have a look.
Thank you very much.
I think all of that is very important, but the thing about 30 minutes is interesting.
The standard life for cell phone relay station batteries is 30 minutes, too.
I wonder if there is any connection.
I think they are changing it to 24 hours at the moment,
so perhaps the battery life of a portion of cell phone relay stations may change a bit.
That thing with 27 minutes was very interesting.
As he was saying, in reality the cell phone batteries held up.
But even in cases where there were generators and cell phone
batteries available, sometimes the relay station had been swept away.
Supposedly the cell phone relay station batteries
lasted for about 24 hours, actually.
After that, probably also due to congestion, communication became impossible.
Of course you're right about that cell phone disaster mode,
but I think the relay stations are pretty important, too.
It might have been good to analyze the traffic data from a mobile operator as well.
Thank you very much.