Marine Data Literacy 2.0

Providing instruction for managing, converting, analyzing and displaying oceanographic station data, marine meteorological data, GIS-compatible marine and coastal data or model simulations, and mapped remote sensing imagery





Home > 5. Gridded Data > 5.3 Resampling Grid

5.3 Resampling Gridded Data to the Project Map Extent with a "Dummy" Grid In Saga

  • Exercise Title:  Resampling Gridded Data to the Project Map Extent with a "Dummy" Grid in Saga

  • Abstract:  "Resampling" is the method to resize or change the resolution of a data grid in GIS systems.  By this means, every aspect of the original grid (columns, rows, cellsize, exact corner locations) can be altered.  This can be done manually, by specifying each of the new parameters, but it is infinitely easier to accomplish with a set (library) of pre-made dummy grids.  A major new project, such as the creation of the Liberia data collection, requires you to make this library, and always use it for resampling, to insure that all project grids are identical.

  • Preliminary Reading (in OceanTeacher, unless otherwise indicated):

    • N/A

  • Required Software:

  • Other Resources: 

  • Author:  Murray Brown

  • Version:  April 2012

1.  Before we can begin gridding, we need to load the "dummy grids" that are needed by most gridding algorithms in Saga as blueprints for the grids to create. 
2.  In Saga, use FILE > GRID > LOAD GRID to load all the dummy grids you have for Liberia.  You should have at least the 1-degree version and the 0.1-degree version.
3.  Also in Saga, use FILE > GRID > LOAD GRID and navigate to the DATA > BASEMAP > GEOLOGY folder to load the global sediment thickness grid, sediments_global_ngdc, you worked with in the Marine GIS exercises.  If you did not do those exercises, then download the zipfile to the same folder location and unzip it.
4.  This is approximately what you should see in Saga, if you have opened a map showing the global sediment grid.
5.  If your map doesn't look quite like the above, it's because we have adjusted the color palette to emphasis "low end" numbers (i.e. we have applied a logarithmic color scale that gives more weight to numbers at the lower end of the range).  The full range of sediment thickness values is about 0 to 18000 m.

You can do this in the properties window, by setting the DISPLAY MODE to LOGARITHMIC (UP).  Also a STRETCH FACTOR must be set, and values from 10-1000 often do a good job.

[Of course, LOGARITHMIC (DOWN) is available for value distributions that need more emphasis at the high end of the range.]

6.  Now to begin the resampling, select MODULES > GRID TOOLS > RESAMPLING
7.  For GRID SYSTEM, drop down the menu and select the system that contains the grid you want to resample.  That is the huge 0.0833333-degree system.
8.  Now make these additional selections:
  • For GRID, choose the only one available, the global sediments grid
  • For PRESERVE DATA TYPE, check the box
  • For TARGET GRID, choose GRID (which means we will name a grid later)

Click OK and this window closes.

9.  This window opens.  For GRID SYSTEM, select the dummy grid system that is closest in resolution to the 0.083333-degree grid to be resampled.

NOTE:  Data "purists" claim that resampling should never be done with a grid that is even slightly smaller in cell-size than the original.  In this case, we are certainly safe.


10.  For GRID, make sure it says CREATE (or it might over-write an existing other file).

Then click OK.

This window closes.

11. The selection of the method depends on the nature of the resampling:
  • If there aren't any no-data areas, then use BICUBIC SPLINE INTERPOLATION
  • If there are no-data areas (as we have here), then;
    •  If you are resampling to a grid cell size that is very close to the original size, then choose NEAREST NEIGHBOR to avoid the creation of false cell values in no-data areas.
    • If you are resampling to a radically different cell size, then use an interpolation method like BICUBIC SPLINE INTERPOLATION, and then (in a second operation) use an appropriate land- or depth-mask to eliminate false cell values that will be created in no-data areas.  Masking is covered in the exercise Masking Methods for Depth- or Height-Limited Grids in Saga

You can see the results of both methods in Panel 13 below, to illustrate the difference.

Then click OK.

12.  A new grid object appears in the 0.1-degree system.  It has a misleading name ("global"), but that is just to help you recognize the original source.

Right-click on the object and select SHOW GRID.

13.  Here is your new grid (on the bottom).  It matches up very well with the World Vector Shoreline from GEBCO.

The improper grid on the top was created with BICUBIC SPLINE INTERPOLATION, and you can easily see that a band of false cells extends onto the land by at least 2 cell-widths.

14.  So that's all there is to it.  With the pre-made dummy grids, you can easily and quickly make identical grids from all sorts of data.  You'll find this necessary many time in the process of making a data collection, because grid operations usually require grids to have specific geometric "systems."  Also, and perhaps even as important, resampled grids are often much smaller than the parent grid, saving storage space, loading time, and calculation time.
15.  Now to save the grids, right-click on it in the objects menu and select SAVE GRID AS.  Then navigate back to the PRODUCTS > SAGA > GRIDS folder and save the grid as sediments_liberia_ngdc_saga_0.1deg.sgrd