With AMRClaw in two space dimensions and GeoClaw it is possible to specify gauge locations as points (x,y) where the values of all components of q should be output every time step during the computation over some time range (t1,t2).

Still need to add to 3d AMRClaw code, and to Classic codes.

Gauges are useful in several ways, e.g.:

  1. To compare computational results to measurements from physical gauges such as a pressure gauge or tide gauge that record data as a function of time at a single point,

  2. To better visualize how the solution behaves at a single point,

  3. To better compare results obtained with different methods or grid resolutions. Comparing two-dimensional pcolor or contour plots can be difficult whereas comparing to curves that give the solution as a function of time often reveals more clearly differences in accuracy or nonphysical oscillations.

Gauge parameters in setrun

See also Specifying AMRClaw run-time parameters in

Gauges are specified in setrun by adding lists of gauge data for each desired gauge to the ClawRunData object rundata.gaugedata.gauges. This is initialized as an empty list and new gauges can be specified by:

rundata.gaugedata.gauges.append([gaugeno, x, y, t1, t2])

with values

  • gaugeno : integer

    the number of this gauge

  • x, y : floats

    the location of this gauge

  • t1, t2 : floats

    the time interval over which gauge data should be output.

During the computation the value of all components of q at all gauge locations will be output to a single file fort.gauge in the output directory. Lines of this file have the form:

gaugeno  level  t  q[0]  q[1] ...  q[meqn-1]

where level is the AMR level used to determine the q values at this time. Internally the finest level available at each gauge is used, with bilinear interpolation to the gauge locations from the 4 nearest cell centers.

If you wish to change what is output at these points, you should copy the library routine dumpgauge.f to your own directory and modify it appropriately.


When doing a restart, previous gauge output is deleted unless you are careful to preserve it. See Output files after a restart.

Plotting tools

Several Python plotting tools are available to plot the gauge data, so you do not have to parse the file fort.gauge yourself.

If you want to read in the data for a particular gauge to manipulate it yourself, you can do, for example:

from import ClawPlotData
plotdata = ClawPlotData()
plotdata.outdir = '_output'   # set to the proper output directory
gaugeno = 1                   # gauge number to examine
g = plotdata.getgauge(gaugeno)


  • g.t is the array of times,

  • g.q is the array of values recorded at the gauges (g.q[m,n] is the m`th variable at time `t[n])

In the setplot Python script you can specify plots that are to be done for each gauge, similar to the manner in which you can specify plots that are to be done for each time frame. For example, to plot the component q[0] at each gauge, include in setplot lines of this form:

plotfigure = plotdata.new_plotfigure(name='q[0] at gauges', figno=300, \

# Set up for axes in this figure:
plotaxes = plotfigure.new_plotaxes()
plotaxes.xlimits = 'auto'
plotaxes.ylimits = [-1.5, 1.5]
plotaxes.title = 'q[0]'

# Plot q[0] as blue line:
plotitem = plotaxes.new_plotitem(plot_type='1d_plot')
plotitem.plot_var = 0
plotitem.plotstyle = 'b-'

Note that plotdata.new_plotfigure is called with type=’each_gauge’ which denotes that this plot is to be produced for each gauge found in (When type is not specified, the default is type=’each_frame’ for time frame data).

If you type:

$ make .plots

then html files will be created for the gauge plots along with the time frame plots and will all show up in the index (usually in _plots/_PlotIndex.html).

When using Iplotclaw to interactively view plots, try:

PLOTCLAW> plotgauge 1

to produce the plot for gauge 1, or simply:

PLOTCLAW> plotgauge

to loop through all gauges. If you rerun the code without re-executing Iplotclaw, you can refresh the gauge data via:

PLOTCLAW> cleargauges

You can of course specify more than one plotitem on each plotaxes if you want. For example to plot the each gauge from the current run as a blue line and the same gauge from some previous run (perhaps with a different grid resolution) as a red line, you could add the following lines to the above example:

# Plot q[0] from previous run as red line:
plotitem = plotaxes.new_plotitem(plot_type='1d_plot')
plotitem.plot_var = 0
plotitem.plotstyle = 'r-'
plotitem.outdir = '_output_from_previous_run'

Plotting gauge locations

It is often convenient to plot the locations of the gauges on pcolor or contour plots each time frame. You can do this as follows, for example:

plotfigure = plotdata.new_plotfigure(name='pcolor', figno=0)
plotaxes = plotfigure.new_plotaxes('pcolor')
plotitem = plotaxes.new_plotitem(plot_type='2d_pcolor')
# set other attributes as desired

def addgauges(current_data):
    from pyclaw.plotters import gaugetools
    gaugetools.plot_gauge_locations(current_data.plotdata, \
         gaugenos='all', format_string='ko', add_labels=True)

plotaxes.afteraxes = addgauges

You can replace gaugenos=’all’ by gaugenos=[1,2] or other list of specific gauges to plot. The format_string above specifies a black dot at each gauge location and add_labels=True means that the gauge number will appear next to each gauge.

If you want more control over this plotting you can of course copy the function plot_gauge_locations from to your file and modify at will.


Several of the examples found in $CLAW/amrclaw/examples/ and $CLAW/geoclaw/examples/ contain the specification of gauges.