probAna is used to derive simple probability maps for a set of simulations for one avalanche track. These maps show for each point in space the probability for a chosen parameter to exceed a given threshold. For example, it is possible to compute the probability map of an avalanche to exceed a peak pressure of 1kPa, but it is also possible to chose other result variables and threshold values.
A set of multiple avalanche simulations is required to generate these maps. The simulations can be generated with :pycom1DFA using a parameter variation, different release-, entrainment- or resistance scenarios. :pyrunScripts.runProbAna gives an example: avalanche simulations for the hockey topography are performed with varying release thickness. A probability map based on peak pressure is generated. The output is a raster file (.asc) with values ranging from 0-1. 0 meaning that no simulation exceeded the threshold in this point in space. 1 on the contrary means that all simulations exceeded the threshold. Details on this function, as for example required inputs can be found in: :pyana4Stats.probAna.
To run - example run scripts -------------------------An example on how to generate probability maps for avalanche simulations performed with :pycom1DFA is given in :pyrunScripts.runProbAna, where for avaHockeyChannel simulations are performed with varying release thickness values ranging from 0.75 to 1.75 meters in steps of 0.05 meters. The resulting simulations are then used to generate the probability map with :pyout3Plot.statsPlots.plotProbMap. There is also the option to filter the simulations further - using the function :pyin3Utils.fileHandlerUtils.getFilterDict which generates a parameter dictionary for filtering according to the filter criteria set in the configuration file (ana4Stats/probAnaCfg.ini) of the :pyana4Stats.probAna function. In order to run this example:
- first go to
AvaFrame/avaframe - copy
ana4Stats/probAnaCfg.initoana4Stats/local_probAnaCfg.ini - uncomment
'FILTER'section inlocal_probAnaCfg.iniand insert filter parameters if you want to first filter simulations run:
python3 runScripts/runProbAna.py
Another example is given in :pyrunScripts.runProbAnaCom1DFA.py. Avalanche simulations are performed with the settings defined in the configuration file of :pycom1DFA and in addition a parameter variation is performed according to the parameters set in ana4Stats/probAnaCfg.ini in the section PROBRUN. The parameters to be varied are set in varParList, the type of variation in variationType (options are: percent, range, rangefromci) and the value of the variation in variationValue. Then there are two sampling strategies to choose from, for performing the parameter variation: (1) a latin hypercube sample of all the parameters to be varied (provided in varParList) is generated and simulations are performed using sets of parameters drawn from this sample. Using sampling strategy (2) all the parameters set in PROBRUN are varied on at a time, i.e. simulations are performed for the standard settings of all parameters, except the one parameter to be varied, subsequently the other variations are performed. One probability map is created for all the different simulations and in case of sampling strategy (2), also one map per parameter that is varied once at a time, is created in addition.
This point-wise probability is expressed by the relative frequency of avalanche peak flow field exceeding a certain threshold for a set of deterministic avalanche simulations derived from a range of input parameters (see HyBeBu2019).
In :pyana4Stats.getStats, functions that help to compute statistical properties of simulation results are gathered. :pyana4Stats.getStats.extractMaxValues can be used to determine the maximum peak values of the simulation results. These values can then be plotted using the functions in :pyout3Plot.statsPlots in order to visualise the statistics of a set of avalanche simulations. For further details on the specific functions, have a look at: :pyana4Stats.getStats.
An example on how to use these statistical functions is given in :pyrunScripts.runStatsExample, where for avaHockeyChannel simulations are performed for two different release area scenarios and the release thickness is varied from 0.75 to 1.75 meters in steps of 0.05 meters. The resulting simulations are then analysed using the :pyana4Stats.getStats.extractMaxValues function and plots are generated using the plotting routines from :pyout3Plot.statsPlots . If in the configuration file ana4Stats/getStats.ini the flag aimec is set to True, additionally an moduleAna3AIMEC:ana3AIMEC: Aimec analysis is performed.
- first go to
AvaFrame/avaframe - copy
ana4Stats/getStats.initoana4Stats/local_getStatsCfg.ini - uncomment
'FILTER'section inana4Stats/local_getStatsCfg.iniand insert filter parameters if you want to first filter simulations run:
python3 runScripts/runStatsExample.py
