The netCDF interface is the new standard for files used in CIV; the NETCDF home page is at http://www.unidata.ucar.edu/packages/netcdf/
The subroutines wr_cdf.f90 and rd_cdf.f90 do the job for the 2D and the 3D algorithms.
This file, which use the extension .nc, store the values in the real space
ie the units are the centimeter and the second.
The velocity evaluation is not on a regular grid due to the local convection
( O1 correction ) V*dt/2 also named "convect flow".
To quickly display a netcdf file under windows or unix , use the following
ncdump file , the option -h limit the display to the headers. Description
. windows binary ;
link for other OS
To use this file with MATLAB, you need to install this
toobox.
We define several 1D arrays : X Y U V C & F; X & Y in the real space (cm and seconds); U & V the velocity ; C the correlation value and F the flag described in the FLAG rules.
2 integers specify the number of coordonates and the space dimensions : nb_coord
& nb_dim for the 3D and/or 3C case.
A byte named constant_pixcm inform about the integration of lenses deformations
or not, if not the real pixcmx & pixcmy defined the ratio between real
and camera space, instead a new array give for each vector the local pixcm
vector.
A real Absolut_time_T0 gives the time elapsed since the first burst.
A integer hart inform about the slection of the hart option during the processing.
A real dt give the time elapsed between the two images used for the computation.
A integerFIX is set to 1 if the field has been fixed
A integerPATCH is set to 1 is the field have been patched
The FIX algorithm add a 1D array named "vec_FixFlag" to identify false vectors ; if this flag is ZERO , the vector is valid.
The thin plate interpolation scheme add severals 1D array to the file. The
interpolation is estimated on a regular grid saved as vec_patch_X and vec_patch_Y
, the corresponding velocities and derivaties are written in the followings
arrays : vec_patch_U vec_patch_V vec_patch_DUDX vec_patch_DUDY vec_patch_DVDX
vec_patch_DVDY. The lenght of these 1D arrays is nb_vec_patch , the filter
cutwave and smooth coefficient are saved as attributes : ro and kc.
All these arrays have twins ( a 0 is added in the named after patch : for example vec_patch0_DVDX) corresponding to an interpolation without filter or smooth coefficient.
Array F gives informations about the CIV1 processing results.
FIX will add a new array FixFlag