Entry Name: "SMU-Betty-MC2"

VAST Challenge 2019
Mini-Challenge 2

Team Members:

Betty Chan Lai Heng, Singapore Management University, betty.chan.2018@mitb.smu.edu.sg PRIMARY
Dr. Kam Tin Seong, Singapore Management University, tskam@smu.edu.sg

Student Team: YES

Tools Used:

Tableau

QGIS

Excel

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

200

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

Video

Included in the folder

Questions

Your task, as supported by visual analytics that you apply, is to help St. Himark's emergency management team combine data from the government-operated stationary monitors with data from citizen-operated mobile sensors to help them better understand conditions in the city and identify likely locations that will require further monitoring, cleanup, or even evacuation. Will data from citizen scientists clarify the situation or make it more uncertain? Use visual analytics to develop responses to the questions below. Novel visualizations of uncertainty are especially interesting for this mini-challenge.

1 – Visualize radiation measurements over time from both static and mobile sensors to identify areas where radiation over background is detected. Characterize changes over time. Limit your response to 6 images and 500 words.

Prelude

From MC 1 shake intensity data, the earthquake and other major events are identified to have started from 8 April. With this, we used 6-7 April static and mobile sensors readings distribution (refer to Figure 1-1a) to identify that 12cpm to 15cpm is the normal background radiation range. Together with reference from http://radiationnetwork.com/ which indicate radiation above 100 cpm is considered in alert state. We will use these values as benchmark in this analysis, where color scale is set as below (refer to Figure 1-1b).

Figure 1-1a: Static and Mobile Sensors Radiation Distributions for 6-7 April

Figure 1-1b: Color Scale

Figure 1-2 depicts the radiation readings at per minute intervals over the 5 days for static and mobile (aggregated) sensors. Each timestep reading is represented by a circle (hollow) shape with the color representing the radiation intensity.

Static sensors

· No radiation in alert state detected over the 5 days.

· Minor spike in Old Town(S9) and Cheddarford on 7 April.

· Old Town detected higher radiation for multiple timesteps from 8-9 April.

· Safe Town (S15), the nearest sensor to Always Safe Nuclear Plant, started to have missing timesteps from 9 April and only resumed readings on late 10 April while S13 detected some minor spikes over these days.

Mobile sensors

· Many neighbourhoods have missing steps, especially those in the South Eastern region such as Oak Willow, Wilson Forest, Scenic Vista and Chapparal (doesn’t have any reading on 10 April).

· Radiation in alert state was detected by Old Town and Safe Town on 8 Apr, and Wilson Forest from 9-10 April.

· Higher radiation was detected by East Parton from 8 April and Scenic Vista from 9 April.

Old Town, Safe Town, Palace Hills and Southwest generally had higher radiation while Downtown and Weston had lower radiation during mid-day, thus created a “ripple effect” on the data points representing these neighbourhoods for the 6-10 April. Static sensors in Old Town had increased marginally while mobile sensors in some neighbourhoods had detected radiation in alert state from 8 April onwards.

Figure 1-2: Static and Mobile Sensors Measurements by Neighborhood for the 5 days

2 – Use visual analytics to represent and analyze uncertainty in the measurement of radiation across the city.

a. Compare uncertainty of the static sensors to the mobile sensors. What anomalies can you see? Are there sensors that are too uncertain to trust?

b. Which regions of the city have greater uncertainty of radiation measurement? Use visual analytics to explain your rationale.

c. What effects do you see in the sensor readings after the earthquake and other major events? What effect do these events have on uncertainty?

Limit your responses to 12 images and 1000 words.

a.

Variation in Readings

Static sensors had negative value readings and mobile sensor had some extreme outliers with maximum readings of 57,345 cpm. Figure 2a-1 shows static and mobile sensors had a wide variation in their readings. Static and mobile sensor had 4 Standard Deviation value of 62 and 689 respectively.

Figure 2a-1: Readings Distribution for Static and Mobile sensors from 6-10 April

Constant Readings (Highlighted in Purple)

M1, M23, M26, M35, and M47 had measured constant readings from 8 April onwards.

High Readings (Highlighted in Pink)

Old Town (S9) and Broadview (S11) had measured high readings from 8 April, and Old Town (S9) also started to measure high readings from 9 April. M13 and M32 had measured multiple timesteps of high readings with some spikes of alert readings.

Alert Readings (Highlighted in Red)

Multiple timesteps of alert readings (i.e. over 100 cpm) were observed after 8 April by some mobile sensors (M9, M10, M20, M21, M22, M24, M25, M27, M28, M29 and M45) and most of them (e.g. M21, M22, M45) had no readings before and after the spike, while static sensors only observed few timesteps of alert readings occasionally.

Below Normal Range Readings (Highlighted in Blue)

M5, M16, M25, M37 and M43 had below normal range readings for some continuous timesteps during the available period.

M5, M16, M25, M37 and M43 with multiple timesteps having constant lower than normal range around the earthquake period are deemed to be too uncertain to trust.

Figure 2a-2: Static and Mobile sensors

b.

Safe Town, Weston and Scenic Vista have greater uncertainty of radiation measurements as conflicting measurements were detected by different sensors for these neighbourhoods.

Safe Town

Some continous/multiple timesteps of extreme radiation readings were measured by some mobile sensors while static sensors only detected occasional spikes of extreme readings and S15 reading was not available from 9 April till end of 10 April.

Figure 2b-1: Safe Town

Weston

Multiple timesteps of below normal and high radiation readings were measured by different mobile sensors resulted in conflicting observations.

Figure 2b-2: Weston

Scenic Vista

M23 measured constant readings in normal range while M20 measured continous extreme readings from 9 April.

Figure 2b-3: Scenic Vista

c.

Interruption in measurements

· M6, M34, M48 and M49 had stopped sending readings from 8 April morning (after the earthquake).

· S15 readings had also stopped from 8 April evening and only resumed on 10 April night.

· Many other mobile sensors (such as M2, M9, M10, M21, M22, M45 and M46) also encountered readings interruptions for couple of timesteps during those events.

These malfunction or connection interruptions occurrences of multiple sensors coincided with the events incidence period thus we inferred that these are the effect of those events.

Figure 2c-1: Static and Mobile Sensors Readings with Interruptions

Effect on Uncertainty

· These effects had caused some neighbourhoods (such as Wilson Forest, Scenic Vista, Chapparal, Terrapin Springs, Pepper Mill and Oak Willow) to loss visibility of the radiation measurements for multiple timesteps as they also do not have static sensors.

· The loss of reading for S15 in Safe Town (which is the nearest static sensor to Always Safe Nuclear Plant) had caused uncertainty on the status of the plant for the reason of the unavailability of the readings, such as power disruption, human intervention to stop readings, etc.

Figure 2c-2: Static and Mobile Sensors by Neighbourhood for 8-10 April

3 – Given the uncertainty you observed in question 2, are the radiation measurements reliable enough to locate areas of concern?

a. Highlight potential locations of contamination, including the locations of contaminated cars. Should St. Himark officials be worried about contaminated cars moving around the city?

b. Estimate how many cars may have been contaminated when coolant leaked from the Always Safe plant. Use visual analysis of radiation measurements to determine if any have left the area.

c. Indicated where you would deploy more sensors to improve radiation monitoring in the city. Would you recommend more static sensors or more mobile sensors or both? Use your visualization of radiation measurement uncertainty to justify your recommendation.

Limit your responses to 10 images and 1000 words

a.

The 11 mobile sensors identified in question 2a (M9, M10, M20, M21, M22, M24, M25, M27, M28, M29 and M45) with multiple timesteps of extreme readings after 8 April are the cars possibly been to the potential contaminated locations.

From the trails of these cars (refer to Figure 3a),

· Old Town, Safe Town, Scenic Vista and Wilson Forest neighbourhoods are identified to be the potential locations of contamination.

· Downtown, Easton, West Parton, Oak Willow, Broadview, Chapparal, Scenic Vista and Terrapin Springs only have a few sparse extreme readings thus they are unlikely the potential contaminated locations.

· The extreme readings detected by theses cars were not consistent throughout all the places they went thus unlikely that these cars “brought the contamintaion” with them around. The high or extreme readings detected at various places likely due to other causes thus the officials do not have to worry about the contaminated cars moving around the city.

Figure 3a-1: Mobile Sensor Trails (Extreme Readings Only) for 8 April to 10 April

Figure 3a-2: M9 Trail (Extreme Readings Only) – Safe Town for 8 April to 10 April [Animated GIF]

Figure 3a-3: M10 Trail (Extreme Readings Only) – Old Town for 8 April to 10 April [Animated GIF]

Figure 3a-4: M20 Trail (Extreme Readings Only) – Scenic Vista for 8 April to 10 April [Animated GIF]

Figure 3a-5: M21, M22, M24, M25, M27, M28, M29 & M45 Trail (Extreme Readings Only) – Wilson Forest for 8 April to 10 April [Animated GIF]

b.

Narrowing to 8 April (earthquake incidence day), M9 and M22 with observed extreme high radiation readings in Safe Town (the neighbourhood that Always Safe Plant is located) are the potential contaminated cars.

Figure 3b-1: Contaminated Cars - Selection

Trail of M9 and M22 from 8-10 April

· Both M9 and M22 left Safe Town.

· M9 had went to Old Town, Southwest, Downtown, Easton, Weston, Southton, East Parton, West Parton.

· M22 had went to Wilson Forest, Scenic Vista, Chapparal, Terrapin Springs, Pepper Mill, Cheddarford.

Figure 3b-2: Contaminated Cars (M9 & M22) – Trail (from 8 Apr to 10 Apr)

Figure 3b-3: Contaminated Cars (M9) – Trail (from 8 Apr to 10 Apr) [Animated GIF]

Figure 3b-4: Contaminated Cars (M22) – Trail (from 8 Apr to 10 Apr) [Animated GIF]

c.

Propose to deploy more static sensors to those neighbourhoods (Wilson Forest, Scenic Vista, Chapparal, Terrapin Springs, Pepper Mill and Oak Willow) identified in question 2c that were affected by the malfunction or connection interruptions of the mobile sensors, so as to avoid such interruption during incidence as we saw that static sensors readings are generally not affected by these events (except S15 in Safe Town which encountered interruption of readings for about 2 days).

More mobile sensor can be deployed to the neighbourhoods that do not have static sensors, nonetheless mobile sensors shall still be deployed to other neighbourhoods to compliance the static sensors readings.

Figure 3c-1: Mobile Sensors Trail for 10 Apr

4 –– Summarize the state of radiation measurements at the end of the available period. Use your novel visualizations and analysis approaches to suggest a course of action for the city. Use visual analytics to compare the static sensor network to the mobile sensor network. What are the strengths and weaknesses of each approach? How do they support each other? Limit your response to 6 images and 800 words.

State of Radiation Measurements at the end of the period

· Multiple mobile sensors had detected extreme readings at Wilson Forest and also multiple skipped timesteps.

· Safe Town did not detect as prolong extreme radiation readings like Wilson Forest despite Always Safe Nuclear Plant is located in this neighbourhood.

· Both static and mobile sensors detected higher radiation in Old Town.

· Mobile sensors detected higher radiation in East Parton and Scenic Vista .

· Most of the neighbourhoods (such as Oak Willow, Terrapin Springs, Pepper Mill, Wilson Forest and Scenic Vista) which do not have static sensors had loss visibility of the radiation measurements for multiple timesteps.

Figure 4-1: Static and Mobile (Aggregated) Reading @ minute interval on 10 April

Propose Course of Actions

· Investigate Wilson Forest for the cause of extreme readings.

· Investigate Safe Town, Weston and Scenic Vista where conflicting measurements were detected by different sensors (as identified in question 2b).

· Investigate the South Eastern region where almost no mobile readings from that area, except M20, M26 and M35 but they were stationary at a location (refer to Figure 4-2) thus they may be leveraging on other available connection channel instead cellphone network.

· Assist or educate citizen scientists on calibrating the mobile sensors to prevent conflicting readings in mobile sensors that cause uncertainty.

Figure 4-2: Mobile Sensors Trail on 10 April

Strength and Weakness of Static vs Mobile network

· Mobile sensor network could be a good complement to static sensor network to provide a complete coverage of the entire city for those neighbourhoods without static sensors (e.g. Wilson Forest and Scenic Vista) or area not near to static sensors.

· Whereas there may be area where car is not accessible, e.g. Wilson Forest Nature Preserve area, where static mobile may be more suitable or forest ranger replace the car (refer to Figure 4.3).

· Static sensors have more stable measurements and higher availability of data (more stable connectivity) than mobile sensors, while mobile sensors are more cost effective to have a more complete coverage of the city.

Figure 4-3: Mobile Sensors Trail (Complete) on 6-10 April – Wilson Forest No Coverage

5 –– 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 analysis is done using the data as a dynamic stream to view the change in measurement over time and quickly be alerted of any crisis or incident. Figure 5-1 illustrates the location of the sensor where extreme reading is dectected and the plot shows the corresponding readings over time. Comparison over time will be difficult if the data is used as a static collection thus difficult to identify crisis or incident and cause the lost in reaction time.

Viewer/User of Figure 5-1 can notice the change in radiation measurement easily on the mid-day of 8 April where dense extreme readings started to appear.

Figure 5-1: Mobile Sensors Trail (Extreme Readings Only, Excluding the extreme outlier 57,345 cpm) for 6-10 April [Animated GIF]

While the data was used as a static collection to derive the normal radiation range for static and mobile sensors. These cut-off values are required to determine the range of radiation for the scales (e.g. color scale) and identify the contaminated place and cars in the analysis. Without the identifying the normal radiation range, we will not be able to visualise the variation or fluctuation of readings that easily and identify those outliers. Following images illustrate the with and without setting the color range for the mobile trail on the map.

Figure 5-2: Mobile Sensors Trail for 6-10 April (with and without setting the color range)