oggm / glacier_simulator Goto Github PK

While programming the AAR I looked over the area calculation for the glacier wide mass balance and saw that I calculated the area and the whole glacier wide mass balance with a wrong model picture in mind. I orientated myself more in how the plot is showing the area than the model intern area is handled.

To make clear what I mean here a small draft when looking from above, the indices show grid points, first row current implementation in the simulator, second row how the model area looks like

So currently the area is calculated using a trapezoidal area between grid points and the mass balance is calculated at the mean elevation of this trapezoid. And the area around the ELA is accounted partly to the accumulation and partly to the ablation area, depending on where the ELA is located relative in the cell.

So my plan is to use now the model point of view and using the model bin_area, and just use the surface_h to determine if the area shoulf count for the accumulation or ablation area (so no area splitting in one cell according to ELA position)

@fmaussion what do you think?

Try to add a stop button by using threads for the model run.

That's a genuine question ;-)

Right now I'm for adding an option to allow for real climate parameters to be used (temp + precip) instead of linear MB, and maybe even transient random runs 🤩

The bin size can be quite large (~ 50) so the bars are visible, and they can get a color (from Red to Blue) scaled to the mass-balance value.

This will need some rearrangement of the internal calculations of the percentages of the bin area (colours in the top right plot).

Timeseries could be stored in an arry even when "Show timeseries" is off. This way, the plots could be displayed after a run - and one could still have a faster run as well.

Here is the image:

https://edu.oggm.org/en/latest/_static/logos/oggm_edu_s_alpha.png

Thanks!

I've updated the README to reflect that:

https://github.com/OGGM/glacier_simulator/blob/master/HOWTO_RELEASE.md

To do at the next release

http://blog.holoviz.org/panel_0.10.0.html#Busy-indicators

One day, it would be cool to have that! (not urgent)

I may understand why this happens, but we need to work around this problem -> the model behavior should be exact same regardless of the "show" step.

Furthermore, the "advance model" button also stops when "equilibrium" is reached. This also shouldn't happen!

By playing along yesterday I came on two ideas to stop a run when it becomes very unstable:

• we could add a wall-clock timer for each 10 years step that would stop the run if a 10 years step needs too long to run. For example:

import signal

def _timeout_handler(signum, frame):
    raise TimeoutError('This run was killed because of timeout')

step_timeout = 10
signal.signal(signal.SIGALRM, _timeout_handler)

while (t_rate > 0.001) and (model.yr < max_calc_years) and (was_close_zero < 5):
        # Start the timer
        signal.alarm(step_timeout)
        model.run_until(model.yr + dyears)
        signal.alarm(0)
        

Or something similar.

• we could stop the simulation when the glacier grows out of the domain. I think that this solution would be enough and would make the complex one above obsolete.

@pat-schmitt , thoughts?

In the current advanced interface the text looks like:

Years to advance model: 150

Would if be to have a changing text so that it reads:

Advance the model for X years

Where "X" is updated with the slider?

With Holoviews 1.14.0 an AttributeError occurs.
This problem is already solved (see holoviz/holoviews#4742).
Until Holoviews 1.14.1 is released I pin the Holoviews version to 1.13.5.

OGGM returns the Mass-Balance in the unit m/yr, and it must be multiplied with rho_ice to get mm w.e./yr.

I will change the unit in the Mass-Balance plot (upper right figure) and multiply the Glacier-wide Mass-Balance (info text lower right) with rho_ice.

@pat-schmitt at your earliest convenience can you add an Uni Bremen logo to the app?

For some experiments it would be very nice to have two buttons in beginner mode:

• reset and run
• advance

would it be possible?

Next to "length", "area", etc.

I really like the app, this is going in the perfect direction!

A few quick thoughts (not much time today):

• the "climate" part should be left for another app. I.e this app focuses on glacier shape and size
• as such, the "simulation time" slider should be removed and all runs should be "run_until_equilibrium"
• is there a way to signify in the app that the glacier is running? I.e. show the current simulation year or volume change or something....
• if there is a way to include the slope of the bed as slider that would be awesome ( maybe this could replace the choice for parabola or trapeze, which is too complex at this stage
• the lower plot (glacier shape) could also be filled with a glacier
• another bedrock profile: wide at the top, narrow at the bottom

cc @pat-schmitt

I'll give you tools to do that, but it would be nice to have the app version displayed. I'll do it for the explorer first then we can add it here.