Entry Name: "UKN-Frauendorf-MC2"

VAST Challenge 2018
Mini-Challenge 2

Team Members:

Giuliana Dehn, University of Konstanz, giuliana.dehn@uni-konstanz.de

Isabel Piljek, University of Konstanz, isabel.piljek@uni-konstanz.de

Jannik Frauendorf, University of Konstanz, jannik.frauendorf@uni-konstanz.de PRIMARY

Ziad Salem, University of Konstanz, ziad.salem@uni-konstanz.de

Yerzhan Niyazbayev, University of Konstanz, yerzhan.niyazbayev@uni-konstanz.de

Juri Buchmüller juri.buchmueller@uni-konstanz.de

Eren Cakmak eren.cakmak@uni-konstanz.de

Wolfgang Jentner wolfgang.jentner@uni-konstanz.de

Florian Stoffel florian.stoffel@uni-konstanz.de

Daniel Keim daniel.keim@uni-konstanz.de

Student Team: YES

Tools Used:

Custom Java Program - PipitSaver (implemented in JavaFX)

PostgreSQL

Tableau

R

KNIME

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

600 hours.

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2018 is complete? YES

Video

https://youtu.be/0ZmvjMNWAZ8 

Link to PipitSaver.jar

https://drive.google.com/open?id=1ljqM7UqfXlOU-CYzNTI0Fdut3k2gP5-G

Questions

1. Characterize the past and most recent situation with respect to chemical contamination in the Boonsong Lekagul waterways. Do you see any trends of possible interest in this investigation? Your submission for this questions should contain no more than 10 images and 1000 words.

Tool description:

The left upper part shows the selected measures which always maps the right upper visualization (crystal view) and right bottom visualization (mosaic view). The crystal view depicts boxplot similar visualizations for each selected measure. They are scaled on a logarithmic base and refer to the two time windows of the timeline in the middle (stream graph). If a section is empty it means that at least one window contains no samples. In order to only plot one time window within the crystal objects the other time window can be deselected (see in Figure 1_1, only window 1 is selected). In the middle of the crystal objects the normalized variances are mapped to grey scales. Dark grey indicates a high variance. The layout of the objects reflects the geographical situation of the measuring stations on the map provided.

On the left side the options for the mosaic view are available to switch the focus of the analysis. One can choose between the number of measurements (grey) and the variances (orange). The values have been aggregated into quarters to provide an overview of the entire period. The values are displayed in the mosaic view on the right side. Sparklines show how the number of measurements changes in relation to the previous quarter - whether these increase, decrease or stay constant.

Figure 1_1

1) Jumps of values at year turns:

Many measures’ values jump over the year turns between 2007 and 2010. We suspect that this can be related to environmental changes such as chemical contaminations. See Figure 1_2: The plotted measures show huge value changes between 2007 and 2008. Especially Atrazine and Endrine, since there is no value range, but thin lines. This means that the values were stable and jumped to new stable values.

Figure 1_2

See Figure 1_3: The plotted measures show huge value changes between 2008 and 2009. The most significant jumps are the decreasing values of Isodrin and Hexachlorobenzene in Kohsoom.

Figure 1_3

Between 2009 and 2010 we also have these value changes: The measure gamma-Hexachlorocyclohexane, which is dangerous for the environment, increased in every location where samples have been taken. It is possible that the measure was distributed over the air so that it reached all locations.

After identifying these huge changes over three years, we checked for more details in a scatter plot (see Figure 1_4) which also shows the jumps at a glance.

Figure 1_4

2) Dumps of Methylosmoline:

We suspect that the dumping of Methyosmoline started in the Christmas break between 2014/2015 when there are no/less Ranger patrols. The dumps were placed near the marked location, see Figure 1_5.

There are different Methyosmoline values per day in Chai; therefore, we assume, that the samples are taken from different rivers 1 and 2 per day in Chai (see Figure 1_5).

Furthermore, We assume that the contamination reached the marked revier 1 directly, resulting in a significant increase of Methyolosmoline in Chai. The contamination also reached Kohsoom (and later also Chai) with revier 2 slowly resulting in a continuously parallel incline of the values. The proved dumps within the year 2015 and 2016 from the last year’s VAST Challenge cause a further increase at Kohsoom.

Moreover, the contamination is more catastrophic than expected. We suspect that Kasios company dumped also near Somchair in the Christmas break 2014/2015 and dumped even a higher amount in the Christmas break 2015/2016.

Figure 1_5

3)  Cover-up of illegal waste dumps:

It is visible at a glance that other measures also show this break within the period of time:

First, the toxic Chlorodinine is also constant for years and decreases to lower concentrations after year turn 2015/2016.

Second, there is an immediate drop in concentration of several other measures.

Third, there is an unusual peak of water temperature recorded at the year turn 2015/2016 - in the middle of the winter!

These patterns lead to the assumption that the water was manipulated. We assume that the Kasios company may diluted the contaminated rivers with clean water in order to conceal the illegal waste dumping.

Macrozoobenthos are indicators for high water quality. They have been sampled in the past, but were unfortunately not sampled in the contaminated locations. So it is not possible to prove the harmful impact of Methyolsomoline.

Figure 1_6 and Figure 1_7 show interesting findings regarding suspicious measures which occured at the end of the sampling period.

Figure 1_6

Figure 1_7

4) Peaks of metal measures:

In the first nine years there were several peaks of certain metals recorded. One of these peaks also contains the highest value in the entire data set (Iron in Kohsoom).

Figure 1_8

5) Correlation:

We used correlation statistic to measure the degree of the relationship between measures. For example, in Achara and Boonsri we found correlation between total hardness and total dissolved salts, and between Sulphates and Sodium and etc. There are correlations between Bicarbonates, Chlorides, Sulphates, and Sodium in Kohsoom and Boonsri. Fecal coliforms and Total coliforms are correlated in Kannika. Iron and Manganese are highly correlated in all four locations.

Figure 1_9

6) Variances:

Figure 1_10 shows chemicals within the mosaic view, displaying the changes of variances over the whole time period. Chemicals like Barium show high variances over the whole time of their occurence. Arsenic show variances starting in 2008 and has its highest variance in Tansanee in 2015. Fecal coliforms and fecal streptococci have the highest variances in Kohsoom and Somchair and Iron has only high variances in one quarter in 2003.

Figure 1_10

2. What anomalies do you find in the waterway samples dataset? How do these affect your analysis of potential problems to the environment? Is the Hydrology Department collecting sufficient data to understand the comprehensive situation across the Preserve? What changes would you propose to make in the sampling approach to best understand the situation? Your submission for this question should contain no more than 6 images and 500 words.

The dataset contains 106 chemicals, from 1998 to 2016 at 10 locations. 32900 measurements are duplicates, were taken on same date and location and has the same value. The dataset covers 28,3% of the whole time range.

Figure 2_1

By looking deeper, in Figure 2_1, measurements were taken in Boonsri, Chai and Kannika (from January 1998), and Busarakham, Kohsoom, Sakda and Somcha

ir (from April 1998). Then the hydrology department improved the sampling through the expansion of three new measurement locations (2009). Furthermore, substances have been measured multiple times on the same date.

Figure 2_2

Between 2005 and 2010, many duplicates, Figure 2_2, have been taken in Boonsri and Chai, which explains the increase in the bottom of this figure. In 2016 the measurements of water temperature (days per week) in Chai increased drastically.

Figure 2-3

Taking a look at Figure 2_3 and 2_4, different patterns appear. The hydrology department does not collect sufficient data, that reflects the comprehensive situation. Chemicals, like Dieldrin and Endrin (2_3A1) have been mainly measured between 2005 and 2007 and in the second half of 2009, containing only one value per location. Aldrin has been measured more often and show also high variances (2_3A2). Barium and Boron has only be sampled over a small time period (2_3B1), but showing unstable behavior (2_3B2). Whereas chemicals like AGOC-3A, Chlorodinine and Methylosmoline only appear since 2014 (2_3C1), also showing high variances (2_3C2).

The sampling of these chemicals is not sufficient, as it does not cover the whole time range. It seems, that the hydrology department discusses the sampling strategy for the following year after each christmas break, as changes mostly appear in the first quarter of a year.

Methylosmoline and AGOC-3A are not measured in the Achara, Decha and Tansanee.

So at first it seemed, that the sampling situation improved by the additional locations, but important measure samples are missing there. Additionally, the missing values and gaps do not allow a continuous time analysis, making it hard to identify some patterns.

Figure 2_4

Sampling accuracy: For certain measures it seems that the values are rounded (1998-2003), resulting in significant constant values. With the begin of 2004, the accuracy is higher. We assume that more accurate sensors are used. The rounding of measures makes an interpretation of the patterns and developments difficult, e.g. detecting periodic data.

Figure 2_5

From Figure 2_6 we can see the 50 strongest rules with regard to in 4 location. This analysis shows that the strategy for taking samples is different based on the location. Within a location the rules are very strong, i.e. the same set of measures are taken together with a regular frequency. For instance in Boonsri, If {Ammonium,Calcium,Chemical Oxygen Demand (Cr),Nitrates,Sulphates,Total dissolved salts and Total phosphorus} are taken, then with 100% confidence the {Orthophosphate-phosphorus} will be also taken.

Figure 2_6

Accordingly, the hydrology department has to improve the sampling strategy. Therefore, we have four proposals that are described in Question 3.

3. After reviewing the data, do any of your findings cause particular concern for the Pipit or other wildlife? Would you suggest any changes in the sampling strategy to better understand the waterways situation in the Preserve? Your submission for this question should contain no more than 6 images and 500 words.

The visualization in Figure 3_1 depicts measures changing dramatically over the whole period of time. We figure that especially these developments can have a big influence on the wildlife of the park because the developments are ongoing processes that do not seem unstoppable. All five globally increasing measures are substances downgrading the water quality. Increasing numbers of visitors to the campsites and the newly built ranger station could be the origin of increasing waste and fecal water. This can have a threatening effect on bird populations and other wild animals. Moreover, the highest value of total coliforms, which can cause serious illness and infection symptoms, is taken in Kohsoom and increases in many locations. Arsenic and Atrazine are also chemicals that are dangerous for the environment and increase over all locations containing samples. Also, all globally decreasing measures can have a huge influence on the wildlife. The question arises whether the wildlife can adapt quickly enough to the changes in the environment. Another reason for the steady change of the measures can be a continuous increase of traffic in the wildlife preserve.

The waste dumping took place between Kohsoom and Boonsri. The chemicals are spreading through the river arms and pollute adjacent areas, and as a result, the habitat of animals shrink. Chemicals can be detected in Somchair and contamination also occurs in Sakda. As these locations show similar behavior in chemical contamination, which we assume is caused by the dumping of Methylosmoline, we accuse the Kasios company of waste dumping in Somchair.

Figure 3_1

To improve the sampling strategy we would suggest four approaches concerning the sampling frequency, redundancies within samples, measure stations and the toxicity of chemicals.

In order to improve the sampling strategy, it would be beneficial to take measures on a regular basis and take samples from all locations on the same date. We identified samplings which have same values that have been taken at same location and day.

Figure 3_2

Chemicals which are detected as stable and which only change over long periods of time can be sampled less frequent than chemicals showing interesting patterns. This saves manpower, money and puts the focus on interesting chemicals.

To understand the origin of contaminations, more measuring stations would be beneficial, especially near suspicious areas. Furthermore, measurement stations are presented in Figure 3_3. M1 is near the dumping area and with this measurement station the impact on the river branches between Boonsri and Kohsoom can be further analyzed. Another interesting location to prove the occurrence of Methylosmoline is the location Somchair. By additional stations M2 and M3, the contamination can be further circumscribed, as it is not clear wether something happened at Somchair or if substances are at the lake above or in previous river branches.

The final proposal is to supply a scale that provides information on how toxic different substances are and in what concentrations they become dangerous. This helps to examine negative impacts on the water quality using Visual Analytics.

Figure 3_3