Solver to interpolate sparse 2D data sets onto the FE mesh.

Interpolation is made using external libraries

  1. nn ( a c code for Natural Neighbours interpolation
  2. csa ( : a c code for cubic spline approximation

The user is invited to get familiar with these libraries by using the command line utilities and test case provided with the libraries

Note the libraries are now hosted at GitHub (above links may be out of date):

  1. nn ( a c code for Natural Neighbours interpolation
  2. csa ( : a c code for cubic spline approximation

data are read from an ASCII file (x,y,variable) or from a netcdf File (if data file has extension .nc)

As this solver depends on several external libraries (nn, csa, netcdf) it is not compiled and included by default in the ElmerIceSolver shared library.

To use the solver :
Update 37f46d0 under elmerice branch
Now you can automatically build, test & compile the solver with cmake.

   -DWITH_ScatteredDataInterpolator:BOOL=TRUE \
   -DCSA_LIBRARY=PATH_TO_libcsa.a \

Old Version

  1. nn and csa libraries must be compiled by the user
  2. Go to the ELMER_TRUNK/elmerice/ScatteredDataInterpolator directory
  3. Edit/Update the Makefile and compile/install the solver

SIF contents

The Solver section looks as

 ! This solver is intended to be used once before simulation to import data sets onto the FE mesh  
  Exec Solver = Before simulation

  Equation = "ScatteredInter"

  procedure = "Scattered2DDataInterpolator" "Scattered2DDataInterpolator"
  Bounding Box dx = Real 1.0e3  
  ! will take only the data points that are within Max/Min mesh corrdinates + the real Value
  ! can be usefull in parallel if all the data are stored in one file
  ! (Default) no bounding box
  CheckNaN = Logical True ! Default True; check is interpolation method gives NaN 
        ! By default replace NaN by nearest available value
        ! We can replace NaNs by Real value with following flag
  !Replace NaN by = Real -9999999 

!!!!! NNI or linear (nn-c library)

 ! Default Sibson interpolation
  Variable 1 = String "ZsNNI"
  Variable 1 data file = File "Rand200.txt"

  Variable 2 = String "ZsNNIW"
  Variable 2 data file = File "Rand200.txt"
  Variable 2 W = Real 0.
! W restricts extrapolation by assigning minimal allowed
!    weight for a vertex (normally "-1" or so; lower
!    values correspond to lower reliability; "0" means
!    no extrapolation)
! Default W=-HUGE(RealNumber);i.e. extrapolation allowed

  Variable 3 = String "ZsNNINS"
  Variable 3 data file = File "Rand200.txt"
  Variable 3 method = String "ns"
 ! method Non-Sibsonian interpolation (nn-c); W can be applied here too

  Variable 4 = String "ZsLin"
  Variable 4 data file = File "Rand200.txt"
  Variable 4 method = String "li"
 ! method linear interpolation (nn-c); W no effect here

  Variable 5 = String "ZsFlight"
  Variable 5 data file = File "FlightLines.txt"

!!!!! Cubic spline (csa-c library)
  Variable 6 = String "ZsCS"
  Variable 6 data file = File "Rand200.txt"
  Variable 6 method = String "cs"
 ! method cubic spline (csa)
  Variable 6 nppc = integer 5
!set the average number of points per cell (default = 5,
!works best for uniform data. Decrease to get smaller
!               cells or increase to get larger cells)
  Variable 6 k = integer 140
! set the spline sensitivity (default = 140, reduce to get
!                     smoother results)


Keywords specific to netcdf:

Solver 2
  Exec Solver = Before simulation

  Equation = "ScatteredInter"

  procedure = "Scattered2DDataInterpolator" "Scattered2DDataInterpolator"

  Variable 1 = File "NetcdfVar"  ! name of the variable in the netcdf file as it is case sensitive use "File"
  Variable 1 data file = File "<NETCDF_FILE>.nc"  ! .nc means that the data file is in netcdf format
  Variable 1 Fill Value = Real ... ! Value meaning No Data (if attribute _FillValue is not found for the variable
  Variable 1 x-dim name = File "x" !name of the x dimension (default: x)
  Variable 1 y-dim name = File "y" !name of the y dimension (default: y)
  Variable 1 x-Var name = File "x" !name of the variable with x-coordinates (default: x)
  Variable 1 y-Var name = File "y" !name of the variable with x-coordinates (default: y)
  !!! keywords for the interpolation are identical. see above


Located under [ELMER_TRUNK]/elmerice/Tests/ScatteredDataInterpolator
You can test with

ctest -L ScatteredDataInterpolator
ctest -L netcdf


A test case is provided under
[ELMER_TRUNK]/elmerice/examples/Test_Scattered2DDataInterpolator .

solvers/scattered.txt · Last modified: 2016/11/19 20:00 by fgillet
CC Attribution-Share Alike 4.0 International Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0