Compute directional derivative or gradient from a grid
gmt grdgradient in_grdfile -Gout_grdfile [ -Aazim[/azim2] ] [ -D[a][c][o][n] ] [ -E[m|s|p]azim/elev[+aambient][+ddiffuse][+pspecular][+sshine] ] [ -N[e|t][amp][+ssigma][+ooffset] ] [ -Qc|r|R [ -Rregion ] [ -Sslopefile ] [ -V[level] ] [ -fflags ] [ -nflags ] [ --PAR=value ]
Note: No space is allowed between the option flag and the associated arguments.
grdgradient may be used to compute the directional derivative in a given direction (-A), or to find the direction (-S) [and the magnitude (-D)] of the vector gradient of the data.
Estimated values in the first/last row/column of output depend on boundary conditions (see -n).
- 2-D grid file from which to compute directional derivative. (See GRID FILE FORMATS below).
- Name of the output grid file for the directional derivative. (See GRID FILE FORMATS below).
- Azimuthal direction for a directional derivative; azim is the angle in the x,y plane measured in degrees positive clockwise from north (the +y direction) toward east (the +x direction). The negative of the directional derivative, -[dz/dx*sin(azim) + dz/dy*cos(azim)], is found; negation yields positive values when the slope of z(x,y) is downhill in the azim direction, the correct sense for shading the illumination of an image (see grdimage and grdview) by a light source above the x,y plane shining from the azim direction. Optionally, supply two azimuths, -Aazim/azim2, in which case the gradients in each of these directions are calculated and the one larger in magnitude is retained; this is useful for illuminating data with two directions of lineated structures, e.g., -A0/270 illuminates from the north (top) and west (left). Finally, if azim is a file it must be a grid of the same domain, spacing and registration as in_grdfile and we will update the azimuth at each output node when computing the directional derivatives.
- Find the direction of the positive (up-slope) gradient of the data. To instead find the aspect (the down-slope direction), use -Da. By default, directions are measured clockwise from north, as azim in -A above. Append c to use conventional Cartesian angles measured counterclockwise from the positive x (east) direction. Append o to report orientations (0-180) rather than directions (0-360). Append n to add 90 degrees to all angles (e.g., to give local strikes of the surface ).
- Compute Lambertian radiance appropriate to use with grdimage and grdview. The Lambertian Reflection assumes an ideal surface that reflects all the light that strikes it and the surface appears equally bright from all viewing directions. Here, azim and elev are the azimuth and elevation of the light vector. Optionally, supply ambient [0.55], diffuse [0.6], specular [0.4], or shine , which are parameters that control the reflectance properties of the surface. Default values are given in the brackets. Use -Es for a simpler Lambertian algorithm. Note that with this form you only have to provide azimuth and elevation. Alternatively, use -Ep for the Peucker piecewise linear approximation (simpler but faster algorithm; in this case the azim and elev are hardwired to 315 and 45 degrees. This means that even if you provide other values they will be ignored.)
- Normalization. [Default is no normalization.] The actual gradients g are offset and scaled to produce normalized gradients gn with a maximum output magnitude of amp. If amp is not given, default amp = 1. If offset is not given, it is set to the average of g. -N yields gn = amp * (g - offset)/max(abs(g - offset)). -Ne normalizes using a cumulative Laplace distribution yielding gn = amp * (1.0 - exp(sqrt(2) * (g - offset)/ sigma)), where sigma is estimated using the L1 norm of (g - offset) if it is not given. -Nt normalizes using a cumulative Cauchy distribution yielding gn = (2 * amp / PI) * atan( (g - offset)/ sigma) where sigma is estimated using the L2 norm of (g - offset) if it is not given. To use offset and/or sigma from a previous calculation, leave out the argument to the modifier(s) and see -Q for usage.
- Controls how normalization via -N is carried out. When multiple grids should be normalized the same way (i.e., with the same offset and/or sigma), we must pass these values via -N. However, this is inconvenient if we compute these values from a grid. Use -Qc to save the results of offset and sigma to a statistics file; if grid output is not needed for this run then do not specify -G. For subsequent runs, just use -Qr to read these values. Using -QR will read then delete the statistics file. See TILES for more information.
- -Rxmin/xmax/ymin/ymax[+r][+uunit] (more …)
- Specify the region of interest. Using the -R option will select a subsection of in_grdfile grid. If this subsection exceeds the boundaries of the grid, only the common region will be extracted.
- Name of output grid file with scalar magnitudes of gradient vectors. Requires -D but makes -G optional.
- -V[level] (more …)
- Select verbosity level [c].
- Geographic grids (dimensions of longitude, latitude) will be converted to meters via a “Flat Earth” approximation using the current ellipsoid parameters.
- -n[b|c|l|n][+a][+bBC][+c][+tthreshold] (more …)
- Select interpolation mode for grids.
- -^ or just -
- Print a short message about the syntax of the command, then exits (NOTE: on Windows just use -).
- -+ or just +
- Print an extensive usage (help) message, including the explanation of any module-specific option (but not the GMT common options), then exits.
- -? or no arguments
- Print a complete usage (help) message, including the explanation of all options, then exits.
- Temporarily override a GMT default setting; repeatable. See gmt.conf for parameters.
Grid Distance Units¶
If the grid does not have meter as the horizontal unit, append +uunit to the input file name to convert from the specified unit to meter. If your grid is geographic, convert distances to meters by supplying -fflags instead.
Usually 255 shades are more than enough for visualization purposes. You can save 75% disk space by appending =nb/a to the output filename out_grdfile.
If you want to make several illuminated maps of subregions of a large data set, and you need the illumination effects to be consistent across all the maps, use the -N option and supply the same value of sigma and offset to grdgradient for each map. A good guess is offset = 0 and sigma found by grdinfo -L2 or -L1 applied to an unnormalized gradient grd.
If you simply need the x- or y-derivatives of the grid, use grdmath.
Grid File Formats¶
By default GMT writes out grid as single precision floats in a COARDS-complaint netCDF file format. However, GMT is able to produce grid files in many other commonly used grid file formats and also facilitates so called “packing” of grids, writing out floating point data as 1- or 2-byte integers. (more …)
For very large datasets (or very large plots) you may need to break the job into multiple tiles. It is then important that the normalization of the intensities are handled the same way for each tile. By default, offset and sigma are recalculated for each tile. Hence, different tiles of the same large grid will compute different offset and sigma values. Thus, the intensity for the same directional slope will be different across the final map. This inconsistency can lead to visible changes in image appearance across tile seams. The way to ensure compatible results is to specify the same offset and sigma via the modifiers to -N. However, if these need to be estimated from the large grid then the -Q option can help: Run grdgradient on the full grid (or as large portion of the grid that your computer can handle) and specify -Qc to create a statistics file with the resulting offset and sigma. Then, for each of your grid tile calculations, give +o and/or +s without arguments to -N and specify -Qr. This option will read the values from the hidden statistics file and use them in the normalization. If you use -QR for the final tile then the statistics file is removed after use.
To make a file for illuminating the data in geoid.nc using exp- normalized gradients in the range [-0.6,0.6] imitating light sources in the north and west directions:
gmt grdgradient geoid.nc -A0/270 -Ggradients.nc=nb/a -Ne0.6 -V
To find the azimuth orientations of seafloor fabric in the file topo.nc:
gmt grdgradient topo.nc -Dno -Gazimuths.nc -V
To determine the offset and sigma suitable for normalizing the intensities from topo.nc, do
gmt grdgradient topo.nc -A30 -Nt0.6 -Qc -V
Without -G, only the hidden statistics file is created and no output grid is written.
To use the previously determined offset and sigma to normalize the intensities in tile_3.nc, do
gmt grdgradient tile_3.nc -A30 -Nt0.6+o+s -Qr -V -Gtile_3_int.nc
Horn, B.K.P., Hill-Shading and the Reflectance Map, Proceedings of the IEEE, Vol. 69, No. 1, January 1981, pp. 14-47. (http://people.csail.mit.edu/bkph/papers/Hill-Shading.pdf)