HASSET  

A probability event tree tool to evaluate future eruptive scenarios using Bayesian Inference. Presented as a plugin for QGIS.

USER MANUAL  

STEFANIA BARTOLINI1, ROSA SOBRADELO1,2, JOAN MARTÍ1

1 Group of volcanology, (SIMGEO-UB) CSIC, Institute of Earth Sciences “Jaume Almera”, Barcelona, Spain 2 Aon Benfield UCL Hazard Centre, Department of Earth Sciences, University College London, London, UK

  2  

HASSET 1.0

H A S S E T   1  1. INTRODUCTION   3  2. INSTALLATION   5  2.1 PREVIOUS REQUIREMENTS   5  3. START HASSET   8  3.1 INPUT DATA   8  3.2 LOCATION AND SIZE   10  3.3 DATA SET TOTAL TIME AND PROBABILITY ESTIMATE TIME INTERVAL   11  3.4 RUN BUTTON   12  4. PROBABILITY AT EACH NODE   13  5. PROBABILITY FOR DIFFERENT SCENARIOS   16  6. MOST LIKELY SCENARIOS   20  REFERENCES   21  CONTACTS   22  

  3  

1. INTRODUCTION  Event tree structures constitute one of the most useful and necessary tools of modern volcanology to assess the volcanic hazard of future eruptive scenarios. It evaluates the most relevant sources of uncertainty in estimating the probability occurrence of a future volcanic event. An event tree is a tree graph representation of events in the form of nodes and branches. Each node represents a step and contains a set of possible branches (outcome for that particular category). The nodes are alternative steps from a general prior event, state, or condition through increasingly specific subsequent events to final outcomes. The nodes are independent and the branches are mutually exclusive and exhaustive, this is, they cannot happen simultaneously and they sum up to one. The objective is to outline all relevant possible outcomes of volcanic unrest at progressively higher degrees of detail and assess the hazard of each scenario by estimating its probability of occurrence within a future time interval. HASSET, Hazard Assessment Event Tree, uses this event tree structure (Fig. 1) to make these estimations based on a statistical methodology called Bayesian Inference. HASSET accounts for the possibility of flank eruptions, as opposed to only central eruptions, geothermal or seismic unrest, as opposed to only magmatic, felsic or mafic composition, as opposed to no composition, relevant volcanic hazards as possible outputs of an eruption, and the distance of that hazard. HASSET introduces the Delta method to approximate the accuracy in the probability estimates, by constructing a one standard deviation variability interval around the expected probability value for each scenario.

Figure 1. HASSET Event Tree Structure

  4  

HASSET is presented as a free software package in the form of a plugin for the open source geographic information system Quantum Gis (QGIS), as a friendly and dynamic graphical user interface (GUI) plugin, which, once properly installed following a few easy steps, creates a new option in the QGIS menu bar called “Volcano”, where the HASSET model is installed. Currently, the software has been developed for Linux (tested on Ubuntu 14.04). The version for Windows OS is under development. HASSET implements the Bayesian Event Tree where the user previously defines a forecasting time interval. The user provides all volcanological data for the analysis, where then HASSET mergers all this information using the Bayesian Event Tree approach.

To do that, a user-friendly interface will guide the user through all the steps:

• To enter all the data for the analysis. • To compute the probability estimates for each branch in the event tree, and

corresponding variability. • To compute the total probability estimate for different scenarios.

                                                           

  5  

2. INSTALLATION  HASSET package is available upon request to the authors or it can be downloaded online at the website of the CSIC Group of Volcanology of Barcelona (http://www.gvb-csic.es) on the “Software & Databases” tab.

 

2.1 PREVIOUS REQUIREMENTS

- Linux (tested on Ubuntu 14.04) operating system

- Matplotlib à http://matplotlib.org/downloads.html   Ubuntu 14.04 users: Open Terminal. Then type: sudo apt-get install python-matplotlib Mac users: http://www.kyngchaos.com/software/python (is recommended to install also the Frameworks: FreeType, GDAL, UnixImageIO à http://www.kyngchaos.com/software/frameworks)

- R à http://cran.r-project.org    Ubuntu 14.04 users: Open Terminal. Then type: sudo apt-get install r-base-core Mac users: Install also the Rpy2 package: http://www.kyngchaos.com/software/python

- QGIS (http://www.qgis.org/en/site/forusers/download.html). We recommend stable version 2.4

2.2 WHERE DO I COPY “volcano_plugin” FOLDER? 1.a Linux users:

- unrar volcano_plugin.rar in /user/.qgis/python/plugins/* To visualize hidden files (any file that begins with a “.”, in our case “.qgis2” on Ubuntu 14.04) open Home Folder and in View select Show Hidden Files (CTRL+H).

1.b Mac users: - unrar volcano_plugin.rar in /user/.qgis/python/plugins/*

To visualize hidden files (any file that begins with a “.”, in our case “.qgis”): a) Open the terminal (found in /Applications/Utilities/) b) Type the following (without quotation marks) to show hidden files: c) “defaults write com.apple.finder AppleShowAllFiles –bool true”

  6  

d) Hit enter e) Type the following (without quotation marks) to restart the Finder: “killall Finder”

5.Hit enter (You can turn hidden files back off by doing the same thing, but switching the “true” to “false” in step 2. (¡¡¡In Mac Os X Leopard change “-bool true” with “YES”!!!))

* If you can’t see in /user/.qgis/python/plugins/ (/user/.qgis2/python/plugins/) the “plugins” folder:

§ Open Qgis § In the menu select Plugins à Manage plugins § In search, type: “Zoom to Point” and activate the plugin. Automatically the “plugins”

folder is created

2. Open QGIS 3. In the menu Plugins à Manage plugins à Simulation Models 4. Now in the menu you can see Volcano (Fig. 2) and run HASSET

  7  

Figure 2. Volcano in QGIS menu

  8  

3. START HASSET  This is the main GUI (Graphical User Interface) in HASSET (Fig. 3)

Figure 3. HASSET main form

The main form is composed by: - The selection of input parameters through the “BROWSER .csv file” button or manually

and their visualization in a table representing the event tree structure - The introduction of the “Dataset” and “Probability estimate Time Windows” - “RUN” button - “INFO HASSET” for an overview of Hasset and the reference paper.

3.1 INPUT DATA First of all the user must select the .csv (comma-separated value) file, where the input data for Past Data, Priori Weight, and Data Weight is selected from a drop down menu (Fig. 4). The “BROWSER .csv file” button allows to visualize the folder in your computer with .csv extension.

  9  

Figure 4. Import .csv file

The format of input data for the model is the following (Fig. 5)

Figure 5. The format of input data

  10  

3.2 LOCATION AND SIZE User has to define the possible locations for the inminent eruption in five different areas. In the example, we have named them as Central, North, South, East and West (Fig. 6). The coverage area for each location would vary for each volcanic system according to topography, surroundings, and/or important topographic barriers which may impose a different level of hazard and risk depending on what side of the volcano the eruption occurs.

Figure 6. Location node Such as the location node, the user must express the size of the eruption in terms of either the volcanic explosive index (VEI), or simply the magnitude of the eruption. This node is grouped in four mutually exclusive and exhaustive categories. In the example, we consider the VEI grater than 5, VEI 4, VEI 3, and VEI less than 2 (Fig. 7).

  11  

Figure 7. Size node

3.3 DATASET AND PROBABILITY ESTIMATE TIME WINDOWS  The Dataset is the range of the volcanic activity analyzed. In our example, we consider a range of eruptions in the last eight thousand years. The Probability estimate Time Windows is the time window range to evaluate the probability to have at least one eruption. For example, our dataset is eight thousand years and we want to estimate the probability of at least one eruption in the next one hundred years, we have eighty time intervals of data for the study of one hundred years each. For each branch we count the number of intervals where at least one event of that type has occurred. For example, out of 80 time intervals, 18 observed an episode of unrest and 62 did not (Fig. 8). So, the number of time intervals is the result of the ratio between the dataset time interval and the probability time interval. This is evaluated automatically and checks if this value corresponds to the sum of the introduced episodes of unrest and no unrest.  

  12  

Figure 8. Number of Time Intervals

3.4 RUN BUTTON  Once all data have been entered user can click the RUN button (Fig. 9) to evaluate the probability and the standard deviation for each branch of all the eight nodes, and displays them in table and graphical format for simplicity.

Figure 9. RUN button

  13  

4. PROBABILITY AT EACH NODE  Once all data is entered, HASSET computes a probability estimate and corresponding variability for each branch of all the eight nodes, and displays them in table (Fig. 10) and graphical format for simplicity.

Figure 10. Unrest node

The pie chart displays graphically these probabilities and user can export the pie chart graph (Fig. 11).

Figure 11. Unrest pie chart

  14  

On the event tree graph we see the node of unrest highlighted in green to show the user at what point of the event tree we are (Fig. 12). The same applies for all of the remaining seven nodes.

Figure 12. Unrest node: info point of the event tree

Furthermore, information about each node is present (Fig. 13).

Figure 13. Info Unrest node

Figure 12 shows an example of how the Unrest tab displays the output on HASSET. We see the initial believes entered for this node in column of Priori and Data weight, and we can see the 80 time windows of which 18 had an episode of unrest and 62 did not. With this data, the

  15  

probability estimate of having at least one unrest in the next time window of 100 years is 23.17%, versus the complement 76.83% of no unrest. Figures 14 and 15 show the results for the Hazards and Distance node, where after observing the data we get that Fallout and Lava flows account for nearly 80% of the total probability estimate of the occurrence of these particular hazards in the next 100 years, while the possibility of any of this hazard affecting an medium/large area is nearly 50%.

Figure 14. Hazards node

Figure 15. Distance node

  16  

5. PROBABILITY FOR DIFFERENT SCENARIOS HASSET computes the total probability estimate for different scenarios. From the window “Scenarios” (Fig. 16) user can select all scenarios of interest to be evaluated by clicking on the desired branch (Fig. 17).

Figure 16. HASSET menu: Scenarios

Figure 17. Scenarios

As we can see in Figure 18, to compute the probability there are five steps: 1. Select a scenario of interest to be evaluated by clicking on the desired branch 2. Push “EVALUATE TOT PROBABILITY OF SELECTED SCENARIO” button. The

relative probability and standard deviation will be evaluated. 3. Refresh the scenario already evaluated and to calculate following scenarios start again

from step 1. User can compare up to five scenarios. 4. To evaluate the total sum of probability for all the scenarios selected, push “TOTAL

SUM” button. 5. To analyze other scenarios, user can delete the previous scenarios with the “Delete

Scenarios” button and start again from step 1.

  17  

Figure 18. Scenarios steps Note that some eruptive scenarios are formed of different combinations, as in Figure 19, where different types of unrest can trigger a magmatic eruption, and so HASSET allows computing and summing up all cases.

  18  

Figure 19. Scenarios

Also, the Hazards node allows to select more than one option (Figures 20, 21), since the same eruption could produce different hazards.

Figure 20. Scenario

  19  

Figure 21. Scenarios

  20  

6. MOST LIKELY SCENARIOS  Most likely scenarios (Fig. 22) is an HASSET option to compute the five most likely scenarios to occur up to a particular node.

Figure 22. Most likely scenarios

The interpretation of these probabilities in terms of significance is down to the personal judgment of the user, unless we have similar volcanoes to compare with. However, we can identify the relative importance of several scenarios by comparing their probabilities of occurrence, providing an important tool for the decision maker to redirect resources and prioritize emergency plans based on whats most likely to occur.

User can visualize the most likely sceanrio for each node just pushing the button associated with that node (Fig. 23).

Figure 23. Most likely scenarios buttons

In Figure 24 we see that the most likely scenarios to happen are magmatic eruptions, mainly of VEI 2 or less, on the North or East sides of the volcano producing lava flows and fallouts of short distance.

  21  

Figure 24. Distance node: most likely scenarios

                                     

  22  

REFERENCES  Sobradelo, R., Bartolini, S., Martí, J.. HASSET: A probability event tree tool to evaluate future eruptive scenarios based on Bayesian Inference. Presented as a plugin for QGIS. Bull. Of Volcanology, 2013, submitted

                                                                                             

  23  

 

CONTACTS    STEFANIA BARTOLINI: mailto:sbartolini@ictja.csic.es ROSA SOBRADELO: mailto:rsobradelo@ictja.csic.es JOAN MARTÍ MOLIST: mailto:jmarti@ictja.csic.es Web GVB: http://www.gvb-csic.es/index_ENG.htm The Institute of the Earth Sciences Jaume Almera (ICTJA) Lluís Solé i Sabarís s/n 08028 Barcelona (Spain) Phone: +34 93 409 54 10