Entry Name: “SMU-Teo-MC1”

VAST Challenge 2019
Mini-Challenge 1

Team Members:

Teo Wei Ting, Singapore Management University, weiting.teo.2018@mitb.smu.edu.sg (Primary Contact)
Dr Kam Tin Seong, Singapore Management University, tskam@smu.edu.sg

Student Team: No

Tools Used: Tableau

Approximately how many hours were spent working on this submission in total? 30 hours

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2019 is complete? Yes

Video

See attached video

Tableau Dashboard

https://public.tableau.com/profile/wei.ting.teo#!/vizhome/VastMiniChallenge1_TWT_final_7Jul/Story1

Questions

1 – Emergency responders will base their initial response on the earthquake shake map. Use visual analytics to determine how their response should change based on damage reports from citizens on the ground. How would you prioritize neighborhoods for response? Which parts of the city are hardest hit? Limit your response to 1000 words and 10 images.

Note: In this analysis, an earthquake is defined as having a shake intensity of 5 and above. Intensity reading levels from the various damage reports are referred to as “damage level” or “damage reading”.

The earthquake took place on 8 Apr 2020 when the most frequently received shake intensity readings spiked to above 5 for the North and North-East side of St. Himark. In contrast, the most commonly received shake intensity ranges only from 0 to 2 during the pre-quake days on 6 – 7 Apr.

The first earthquake took place on 8 Apr 8.30a.m. The citizens from Old Town, Easton and East Parton were the first to submit the shake reports, with a minimum number of 100 reports submitted.

The first three hours after the earthquake is the most critical period for rescue efforts. It is also as known as the “golden hour” – the period following a traumatic experience during which there is the highest likelihood of preventing casualty, should prompt rescue and/or medical treatments be rendered.

To identify the hardest hit areas, the following factors were considered:

1. The highest damage level received;

2. The damage level of the topmost frequently received readings;

3. The number of reports received.

Golden Hour on 8 Apr 2020 9a.m to 12p.m

Shake Intensity Report

Analysing the shake intensities of 7 and above, we found that the reports were mainly submitted by Old Town and Safe Town, with more than 100 reports submitted by each neighbourhood. This makes it credible that there was indeed a major shake experienced by the citizens on the ground. Hence, these two neighbourhoods experienced the highest shake intensity and are the closest to the epicentre of the earthquake.

As the highest shake intensity might not be felt/reported by many of the citizens of each neighbourhood, the topmost frequently reported shake intensity was also considered. The figure below confirmed that Old Town and Safe Town had the highest intensity level (level 6) across the city during the golden hour.

Other Damage Reports

Though it is established that Old Town and Safe Town are nearest to the epicentre, it is also important to examine the damage level of the various lifeline facilities across St. Himark as they provide critical services which are essential to the well-being of the community e.g. water and sanitation, and electricity etc. Here, we considered the number of serious reports (with high damage level rated 7 and above) received from each neighbourhood on Day 1 of the earthquake.

Note that the damages reported by some neighbourhoods might not be caused by the earthquake directly as i) the neighbourhoods were located too far from the epicentre (near Old Town and Safe Town) and/or ii) some of the lifeline facilities were already old/broken prior to the earthquake and were undergoing repair/maintenance.

Sewer & Water: Broadview and Scenic Vista had submitted the greatest number of damage reports and they have the highest number of serious reports. However, these neighbourhoods are located too far away from the epicentre. Hence, the damages could not have been caused by the earthquake.

Easton and Old Town had also submitted many serious reports. For Old Town, even though there were ongoing repair works to replace the old water supply lines throughout the neighbourhood even before the earthquake (see that there were serious reports sent on 7 Apr), there was a huge spike in the number of serious damage reports submitted on 8 Apr.

Conducting similar analysis using the Buildings and Power damage reports, it is clear that the earthquake had caused the worst damage to these facilities at East Parton, Easton and Old Town. It is not surprising as these neighbourhoods were located near the epicentre.

Roads & Bridges: Prior to the earthquake, Broadview and Scenic Vista were undergoing resurfacing of residential streets. Hence, the serious damage readings could be due to the ongoing construction projects. Besides Easton and Old Town which submitted many serious reports, Northwest had also submitted numerous damage reports of moderate severity (rated 5).

Medical: Old Town had the worst damage to the medical facilities caused by the earthquake.

In conclusion, combining the shake intensity reports and the various damage reports received during the golden hour, the emergency responders should prioritise the neighbourhoods nearest to the epicentre as they have been the most affected by the quake i.e. Old Town, Easton and East Parton. In particular, the Old Town Hospital could provide valuable medical assistance to those who were injured during the earthquake. Easton should also be prioritised as it is a young neighbourhood filled with children who are at greater risk, more vulnerable to stress that follows a natural disaster and require more assistance during crisis. Moreover, Easton contains schools which could be used as temporary relief centres to provide shelter, food and medical supplies to the homeless.

Though Safe Town has reported high shake intensity, the damage readings were not serious. Still, the emergency management team should prioritise Safe Town to secure the Always Safe nuclear power plant as it is the primary power source of the city. Besides, Safe Town contained the other local industries and the damage to the economic power of the city should be minimised.

Lastly, emergency responders should also repair the damages to the roads and bridges at Northwest. This would ensure that evacuation was possible, and the route would be clear for any food/medical supplies to reach the neighbourhoods.

Aftershocks on 9 and 10 Apr 2020

There were aftershocks occurring along the North and Eastern part of the city. Old Town continued to submit the highest shake intensity.

There were some mild aftershocks of ratings 4 to 5 from Old Town, Safe Town and Pepper Mill.

2 – Use visual analytics to show uncertainty in the data. Compare the reliability of neighborhood reports. Which neighborhoods are providing reliable reports? Provide a rationale for your response. Limit your response to 1000 words and 10 images.

Uncertainty can be analysed in terms of:

i) The distribution of data and the number of records provided;

ii) Comparison of the data distribution and damage levels before and after the earthquake;

iii) Comparison of shake intensity vs the damage reports received; and

iv) Comparison of shake intensity report and the shake map.

i) The distribution of data and the number of records provided:

Looking at the shake intensity reports received on the day of the first earthquake on 8 Apr, the uncertainty in the reports was the greatest for Old Town and Safe Town, where the spread/range of the readings were the highest, ranging from 0 to 9 within a day. The interquartile range (IQR) for both towns were 5 (7 - 2).

We could also analyse uncertainty by looking at the number of data points – a lower number of data points imply greater uncertainty in the distribution of the data. Since Safe Town had less data points (as seen from the size of the bubbles), the reports from Safe Town were less reliable than Old Town. Using the filter function, we could also perform similar analysis with the different types of damage reports collected. On the other hand, Northwest and Scenic Vista provided reliable shake intensity reports as we can see from the low IQR and the numerous number of reports submitted.

If we were to select a particular hour and time of the day, we could also see that even within the same time frame in each neighbourhood, there was variation and uncertainty in the data reported.

ii) Comparison of the data distribution and damage levels before and after the earthquake

Comparing the shake intensity reports received before, during and after the earthquake, the reports received before the earthquake (on 6 Apr) were more reliable as shown by the lower spread of the data distribution. The greatest uncertainty came from Old Town and Safe Town, with a range from 0 to 3, and the IQR is 2. Even on the days without the earthquake, Old Town and Safe Town sent some reports with shake intensity readings of 3, indicating uncertainty in the data. This could be due to ongoing construction and repair works which contributed to earth movement records.

The reports received across the city on 10 Apr after the mainshock showed less variability (i.e. less uncertainty), than the reports received on the day of the mainshock on 8 Apr. The reports from Old Town were the least reliable as it has the highest IQR.

There is uncertainty in the damage reports submitted even before the earthquake. For instance, for the Sewer and Water damage reports submitted on 7 Apr, we would not be able to tell if the serious damages were caused by the earthquake or if had already existed before the earthquake. The high IQR of 5 across all neighbourhoods also indicate uncertainty/unreliability in the data.

iii) Comparison of shake intensity vs the damage reports received

We would expect the level of damage reported to correspond according to the shake intensity (i.e. the higher the shake intensity, the higher the level of damages done to the facilities). However, there was uncertainty when we compared the citizens’ reports of shake intensity vs various damage reports. The neighbourhoods which provided unreliable reports were Broadview and Scenic Vista.

During the earthquake on 8 Apr, the most frequently received shake report from Broadview was of intensity 1, which was reasonable since Broadview is far from the epicentre. However, the readings from the Sewer & Water damage report was very high, with a mode of 7.

The Sewer & Water report was not the only unreliable report from Broadview. Using the filter function, it was found that the other damage reports such as the Roads & Bridges, Power and Buildings from Broadview were also unreliable, since it showed very high level of damage, for a low level of shake intensity.

Similarly, Scenic Vista was another neighbourhood which provided data with uncertainty.

In contrast, the Power damage reports received from Old Town are quite reliable as the damage levels correspond with shake intensities.

iv) Comparison of shake intensity report and the shake map

The shake map at the bottom left corner has shown the neighbourhoods experiencing a light to moderate shaking. This seemed to have corresponded with the neighbourhoods identified from citizens’ reports received, as highlighted in the shp. file above.

However, there was uncertainty as to the actual shake intensity felt in these neighbourhoods - the shake map indicated a light shake of 4, yet the citizen’s reports showed a range of shake intensities ranging from 0 to 9. For instance, the most commonly received shake intensity from Old Town was 6, which is higher than what was shown on the shake map. It is also observed that the neighbourhoods closest to the epicentre show higher variation/ uncertainty in the shake intensities sent from the citizens, as shown by their higher IQR.

East Parton and Easton were indicated on the shake map to experience weak shake of intensity 2 to 3. This is another source of uncertainty the most commonly received reports from the citizens of these neighbourhoods were a shake intensity of 4 instead.

The other neighbourhoods shown on the box plot with a lower variability from 0 to 4 shake intensity seem to be more in line with the readings shown on the shake map, with weak or no shake.

3 – How do conditions change over time? How does uncertainty in change over time? Describe the key changes you see. Limit your response to 500 words and 8 images.

In terms of shake intensity across the whole city, the number of reports received (look at the size of the bubbles) peaked on 8 Apr, and gradually declined over the next two days. The intensity of the shakes also reduced as we can see that the number of reports with a shake intensity of 5 and above fell after 8 Apr. The uncertainty/variation in the data is also reduced as the IQR of shake intensities reported fell from 5 on 8 Apr, to 4 on 10 Apr.

This general observation could be seen for the neighbourhoods near the epicentre (see Safe Town below), as well as those which were not located near it (see Terrapin Springs).

One exception was Old Town. The number of shake intensity reports received increased on 9 Apr. These were likely to be the reports of the mild after shakes felt, as the shake intensities reported were of a lower range from 0 to 3, while the number of reports of the higher readings from 5 and above fell on 9 Apr. On 10 Apr, there was some light to moderate after shakes which were only felt in Old Town and not any other neighbourhoods.

Analysing the Sewer and Water damage reports received from the entire city, though the total number of received reports and the number of serious reports declined after the main quake on 8 Apr, the level of data variability did not reduce in terms of the IQR of the damage levels reported.

Zooming in specifically to Old Town, the number of serious reports continued to increase after 8 Apr, where the number of level 10 damage spiked on 10 Apr. This could be due to the downtime due to the repair works which were taking place right after the earthquake, as the emergency management tries to restore the facilities to provide the critical lifeline facilities to the community as soon as possible.

The other damage reports showed similar trends to what was observed for the Sewer and Water report (see example of Roads and Bridges damage report for Old Town).

4 –– The data for this challenge can be analyzed either as a static collection or as a dynamic stream of data, as it would occur in a real emergency. Describe how you analyzed the data - as a static collection or a stream. How do you think this choice affected your analysis? Limit your response to 200 words and 3 images.

The dashboards designed for this challenge has analysed the data as a static collection as all the data points from the entire city, over the entire time period from 6 to 11 Apr were analysed together. If the data had been analysed dynamically, the below dashboard would be used – the damage level of the latest received report would be compared against the reports which were received earlier.

For instance, to facilitate the emergency response team’s decision whether to deploy their resources to say, Old Town on 8 Apr at 3p.m, the historical data would be used as a backdrop i.e. a basis of comparison to determine whether the latest report received warranted any assistance. The response team could look at the number and severity of the reports received from that neighbourhood over the last few hours (in 3-hour time interval shown in this example), and also see how it compares against other neighbourhoods to prioritise the limited resources.