3 A Priori Analogues

Investigating a priori analogue volcanoes

PyVOLCANS can also be used to investigate sets of analogue volcanoes that have been derived using other approaches different from PyVOLCANS, e.g. analogue volcanoes based on expert knowledge. Such analogue volcanoes might be called 'a priori analogues' (Tierz et al., 2019). PyVOLCANS offers the user the possibility of checking for the proportion (or percentage) of Holocene volcanoes in the GVP database that are 'better analogues' (i.e. have a higher value of total analogy with the target volcano), compared to each of the a priori analogues provided by the user.

For example, if we choose Volcán de Fuego (Guatemala) and the following a priori analogues (please see Figure 6 in Tierz et al., 2019): Villarrica, Llaima (Chile), Pacaya (Guatemala), Reventador, Tungurahua (Ecuador), the results are the following:

$ pyvolcans Fuego --apriori Villarrica Llaima Pacaya Reventador Tungurahua

Top 10 analogue volcanoes for Fuego, Guatemala (342090):
         name           country  smithsonian_id  total_analogy
 Klyuchevskoy            Russia          300260       0.970910
       Semeru         Indonesia          263300       0.963634
       Osorno             Chile          358010       0.955407
      Merbabu         Indonesia          263240       0.953673
       Tacana  Mexico-Guatemala          341130       0.951295
  Chikurachki            Russia          290360       0.951209
       Pavlof     United States          312030       0.950073
        Baker     United States          321010       0.949815
   Acatenango         Guatemala          342080       0.949274
   Shishaldin     United States          311360       0.948061


According to PyVOLCANS, the following percentage of volcanoes in the GVP database
are better analogues to Fuego than the 'a priori' analogues reported below:

Villarrica (357120): 2%

Llaima (357110): 2%

Pacaya (342110): 8%

Reventador (352010): 7%

Tungurahua (352080): 7%

Please note that: (1) the percentages are calculated to the closest unit percentage, and (2) given the total number of Holocene volcanoes in the GVP database v.4.6.7 used by PyVOLCANS (N = 1439), 1% corresponds to 14 volcanoes, approximately.

Dependence of better analogues on the weighting scheme

It is also critical to be aware that any value of total analogy calculated by PyVOLCANS, and therefore any percentage of 'better analogues', is not only dependent on the choice of target volcano and a priori analogues, but also, importantly, on the specific choice of weighting scheme used for each run of PyVOLCANS. Different weighting schemes may lead to different sets of top analogue volcanoes as well as to different percentages of 'better analogues' for any pair of target volcano-a priori analogue.

Hence, using the equal-weight scheme for eruption size and style mentioned above (scheme B in Tierz et al., 2019) does modify the PyVOLCANS results in the example for Volcán de Fuego:

$ pyvolcans Fuego -Sz 1/2 -St 1/2 --apriori Villarrica Llaima Pacaya Reventador Tungurahua

Top 10 analogue volcanoes for Fuego, Guatemala (342090):
         name        country  smithsonian_id  total_analogy
    Momotombo      Nicaragua          344090       0.985340
        Pagan  United States          284170       0.982429
       Pavlof  United States          312030       0.982149
 Klyuchevskoy         Russia          300260       0.981609
  Karangetang      Indonesia          267020       0.977516
   Villarrica          Chile          357120       0.977199
       Semeru      Indonesia          263300       0.976578
     Lewotobi      Indonesia          264180       0.976012
     Karymsky         Russia          300130       0.975861
       Ambrym        Vanuatu          257040       0.974410

According to PyVOLCANS, the following percentage of volcanoes in the GVP database
are better analogues to Fuego than the 'a priori' analogues reported below:

Villarrica (357120): 1%

Llaima (357110): 2%

Pacaya (342110): 2%

Reventador (352010): 7%

Tungurahua (352080): 26%

WARNING! Please note that, in the current version of PyVOLCANS, defining the set of a priori analogues as a mix of data types (e.g. str for volcano names and int for volcano number, VNUM (or Smithsonian ID)) does not provide the user with the expected data for the percentages of better analogues.

PLEASE USE either a set of volcano names or a set of volcano numbers to define your set of a priori analogues when running PyVOLCANS. Many thanks.

Plotting capabilities

From PyVOLCANS v1.1.0 onwards, users can also visualise: (1) the single-criterion and total (multi-criteria) analogy values between the target volcano and any a priori analogue volcano selected by the user, and (2) the percentage of better analogues (than each of the a priori analogues) available in the whole GVP database, for the specific target volcano chosen.

These results are provided as bar plots when the user selects the optional flag --plot_apriori (or -pa):

$ pyvolcans Fuego --apriori Villarrica Llaima Pacaya Reventador Tungurahua --plot_apriori

If the user wants to save the generated figures (.png format, 600 dpi resolution), the optional flag --save_figures (or -S) should be selected, together with the --plot_apriori (or -pa) flag.